20220426.md

ArXiv cs.CV --Tue, 26 Apr 2022

1.Proto2Proto: Can you recognize the car, the way I do? ⬇️

Prototypical methods have recently gained a lot of attention due to their intrinsic interpretable nature, which is obtained through the prototypes. With growing use cases of model reuse and distillation, there is a need to also study transfer of interpretability from one model to another. We present Proto2Proto, a novel method to transfer interpretability of one prototypical part network to another via knowledge distillation. Our approach aims to add interpretability to the "dark" knowledge transferred from the teacher to the shallower student model. We propose two novel losses: "Global Explanation" loss and "Patch-Prototype Correspondence" loss to facilitate such a transfer. Global Explanation loss forces the student prototypes to be close to teacher prototypes, and Patch-Prototype Correspondence loss enforces the local representations of the student to be similar to that of the teacher. Further, we propose three novel metrics to evaluate the student's proximity to the teacher as measures of interpretability transfer in our settings. We qualitatively and quantitatively demonstrate the effectiveness of our method on CUB-200-2011 and Stanford Cars datasets. Our experiments show that the proposed method indeed achieves interpretability transfer from teacher to student while simultaneously exhibiting competitive performance.

2.Retrieval-Augmented Diffusion Models ⬇️

Generative image synthesis with diffusion models has recently achieved excellent visual quality in several tasks such as text-based or class-conditional image synthesis. Much of this success is due to a dramatic increase in the computational capacity invested in training these models. This work presents an alternative approach: inspired by its successful application in natural language processing, we propose to complement the diffusion model with a retrieval-based approach and to introduce an explicit memory in the form of an external database. During training, our diffusion model is trained with similar visual features retrieved via CLIP and from the neighborhood of each training instance. By leveraging CLIP's joint image-text embedding space, our model achieves highly competitive performance on tasks for which it has not been explicitly trained, such as class-conditional or text-image synthesis, and can be conditioned on both text and image embeddings. Moreover, we can apply our approach to unconditional generation, where it achieves state-of-the-art performance. Our approach incurs low computational and memory overheads and is easy to implement. We discuss its relationship to concurrent work and will publish code and pretrained models soon.

3.StyleGAN-Human: A Data-Centric Odyssey of Human Generation ⬇️

Unconditional human image generation is an important task in vision and graphics, which enables various applications in the creative industry. Existing studies in this field mainly focus on "network engineering" such as designing new components and objective functions. This work takes a data-centric perspective and investigates multiple critical aspects in "data engineering", which we believe would complement the current practice. To facilitate a comprehensive study, we collect and annotate a large-scale human image dataset with over 230K samples capturing diverse poses and textures. Equipped with this large dataset, we rigorously investigate three essential factors in data engineering for StyleGAN-based human generation, namely data size, data distribution, and data alignment. Extensive experiments reveal several valuable observations w.r.t. these aspects: 1) Large-scale data, more than 40K images, are needed to train a high-fidelity unconditional human generation model with vanilla StyleGAN. 2) A balanced training set helps improve the generation quality with rare face poses compared to the long-tailed counterpart, whereas simply balancing the clothing texture distribution does not effectively bring an improvement. 3) Human GAN models with body centers for alignment outperform models trained using face centers or pelvis points as alignment anchors. In addition, a model zoo and human editing applications are demonstrated to facilitate future research in the community.

4.Zero-Shot Logit Adjustment ⬇️

Semantic-descriptor-based Generalized Zero-Shot Learning (GZSL) poses challenges in recognizing the novel classes in the test phase. The development of generative models enables current GZSL techniques to probe further into the semantic-visual link, culminating in a two-stage form that includes a generator and a classifier. However, existing generation-based methods focus on enhancing the generator's effect while neglecting the improvement of the classifier. In this paper, we first conduct an analysis of two properties of the generated pseudo unseen sample: bias and homogeneity. Then, we perform variational Bayesian inference to back-derive the evaluation metrics, which reflects the balance of the seen and unseen classes. As a consequence of our derivation, the aforementioned two properties are incorporated into the classifier training as seen-unseen priors via logit adjustment. The Zero-Shot Logit Adjustment further puts semantic-based classifiers into effect in generation-based GZSL. Our experiments demonstrate that the proposed technique achieves the state of the art when combined with the basic generator, and it can improve various generative zero-shot learning frameworks. Our codes are available on \url{this https URL}.

5.Real-Time Neural Character Rendering with Pose-Guided Multiplane Images ⬇️

We propose pose-guided multiplane image (MPI) synthesis which can render an animatable character in real scenes with photorealistic quality. We use a portable camera rig to capture the multi-view images along with the driving signal for the moving subject. Our method generalizes the image-to-image translation paradigm, which translates the human pose to a 3D scene representation -- MPIs that can be rendered in free viewpoints, using the multi-views captures as supervision. To fully cultivate the potential of MPI, we propose depth-adaptive MPI which can be learned using variable exposure images while being robust to inaccurate camera registration. Our method demonstrates advantageous novel-view synthesis quality over the state-of-the-art approaches for characters with challenging motions. Moreover, the proposed method is generalizable to novel combinations of training poses and can be explicitly controlled. Our method achieves such expressive and animatable character rendering all in real time, serving as a promising solution for practical applications.

6.Generalizable Neural Performer: Learning Robust Radiance Fields for Human Novel View Synthesis ⬇️

This work targets at using a general deep learning framework to synthesize free-viewpoint images of arbitrary human performers, only requiring a sparse number of camera views as inputs and skirting per-case fine-tuning. The large variation of geometry and appearance, caused by articulated body poses, shapes and clothing types, are the key bottlenecks of this task. To overcome these challenges, we present a simple yet powerful framework, named Generalizable Neural Performer (GNR), that learns a generalizable and robust neural body representation over various geometry and appearance. Specifically, we compress the light fields for novel view human rendering as conditional implicit neural radiance fields from both geometry and appearance aspects. We first introduce an Implicit Geometric Body Embedding strategy to enhance the robustness based on both parametric 3D human body model and multi-view images hints. We further propose a Screen-Space Occlusion-Aware Appearance Blending technique to preserve the high-quality appearance, through interpolating source view appearance to the radiance fields with a relax but approximate geometric guidance.
To evaluate our method, we present our ongoing effort of constructing a dataset with remarkable complexity and diversity. The dataset GeneBody-1.0, includes over 360M frames of 370 subjects under multi-view cameras capturing, performing a large variety of pose actions, along with diverse body shapes, clothing, accessories and hairdos. Experiments on GeneBody-1.0 and ZJU-Mocap show better robustness of our methods than recent state-of-the-art generalizable methods among all cross-dataset, unseen subjects and unseen poses settings. We also demonstrate the competitiveness of our model compared with cutting-edge case-specific ones. Dataset, code and model will be made publicly available.

7.PVNAS: 3D Neural Architecture Search with Point-Voxel Convolution ⬇️

3D neural networks are widely used in real-world applications (e.g., AR/VR headsets, self-driving cars). They are required to be fast and accurate; however, limited hardware resources on edge devices make these requirements rather challenging. Previous work processes 3D data using either voxel-based or point-based neural networks, but both types of 3D models are not hardware-efficient due to the large memory footprint and random memory access. In this paper, we study 3D deep learning from the efficiency perspective. We first systematically analyze the bottlenecks of previous 3D methods. We then combine the best from point-based and voxel-based models together and propose a novel hardware-efficient 3D primitive, Point-Voxel Convolution (PVConv). We further enhance this primitive with the sparse convolution to make it more effective in processing large (outdoor) scenes. Based on our designed 3D primitive, we introduce 3D Neural Architecture Search (3D-NAS) to explore the best 3D network architecture given a resource constraint. We evaluate our proposed method on six representative benchmark datasets, achieving state-of-the-art performance with 1.8-23.7x measured speedup. Furthermore, our method has been deployed to the autonomous racing vehicle of MIT Driverless, achieving larger detection range, higher accuracy and lower latency.

8.Adversarial Attention for Human Motion Synthesis ⬇️

Analysing human motions is a core topic of interest for many disciplines, from Human-Computer Interaction, to entertainment, Virtual Reality and healthcare. Deep learning has achieved impressive results in capturing human pose in real-time. On the other hand, due to high inter-subject variability, human motion analysis models often suffer from not being able to generalise to data from unseen subjects due to very limited specialised datasets available in fields such as healthcare. However, acquiring human motion datasets is highly time-consuming, challenging, and expensive. Hence, human motion synthesis is a crucial research problem within deep learning and computer vision. We present a novel method for controllable human motion synthesis by applying attention-based probabilistic deep adversarial models with end-to-end training. We show that we can generate synthetic human motion over both short- and long-time horizons through the use of adversarial attention. Furthermore, we show that we can improve the classification performance of deep learning models in cases where there is inadequate real data, by supplementing existing datasets with synthetic motions.

9.4DAC: Learning Attribute Compression for Dynamic Point Clouds ⬇️

With the development of the 3D data acquisition facilities, the increasing scale of acquired 3D point clouds poses a challenge to the existing data compression techniques. Although promising performance has been achieved in static point cloud compression, it remains under-explored and challenging to leverage temporal correlations within a point cloud sequence for effective dynamic point cloud compression. In this paper, we study the attribute (e.g., color) compression of dynamic point clouds and present a learning-based framework, termed 4DAC. To reduce temporal redundancy within data, we first build the 3D motion estimation and motion compensation modules with deep neural networks. Then, the attribute residuals produced by the motion compensation component are encoded by the region adaptive hierarchical transform into residual coefficients. In addition, we also propose a deep conditional entropy model to estimate the probability distribution of the transformed coefficients, by incorporating temporal context from consecutive point clouds and the motion estimation/compensation modules. Finally, the data stream is losslessly entropy coded with the predicted distribution. Extensive experiments on several public datasets demonstrate the superior compression performance of the proposed approach.

10.Masked Image Modeling Advances 3D Medical Image Analysis ⬇️

Recently, masked image modeling (MIM) has gained considerable attention due to its capacity to learn from vast amounts of unlabeled data and has been demonstrated to be effective on a wide variety of vision tasks involving natural images. Meanwhile, the potential of self-supervised learning in modeling 3D medical images is anticipated to be immense due to the high quantities of unlabeled images, and the expense and difficulty of quality labels. However, MIM's applicability to medical images remains uncertain. In this paper, we demonstrate that masked image modeling approaches can also advance 3D medical images analysis in addition to natural images. We study how masked image modeling strategies leverage performance from the viewpoints of 3D medical image segmentation as a representative downstream task: i) when compared to naive contrastive learning, masked image modeling approaches accelerate the convergence of supervised training even faster (1.40$\times$) and ultimately produce a higher dice score; ii) predicting raw voxel values with a high masking ratio and a relatively smaller patch size is non-trivial self-supervised pretext-task for medical images modeling; iii) a lightweight decoder or projection head design for reconstruction is powerful for masked image modeling on 3D medical images which speeds up training and reduce cost; iv) finally, we also investigate the effectiveness of MIM methods under different practical scenarios where different image resolutions and labeled data ratios are applied.

11.Tac2Pose: Tactile Object Pose Estimation from the First Touch ⬇️

In this paper, we present Tac2Pose, an object-specific approach to tactile pose estimation from the first touch for known objects. Given the object geometry, we learn a tailored perception model in simulation that estimates a probability distribution over possible object poses given a tactile observation. To do so, we simulate the contact shapes that a dense set of object poses would produce on the sensor. Then, given a new contact shape obtained from the sensor, we match it against the pre-computed set using an object-specific embedding learned using contrastive learning. We obtain contact shapes from the sensor with an object-agnostic calibration step that maps RGB tactile observations to binary contact shapes. This mapping, which can be reused across object and sensor instances, is the only step trained with real sensor data. This results in a perception model that localizes objects from the first real tactile observation. Importantly, it produces pose distributions and can incorporate additional pose constraints coming from other perception systems, contacts, or priors.
We provide quantitative results for 20 objects. Tac2Pose provides high accuracy pose estimations from distinctive tactile observations while regressing meaningful pose distributions to account for those contact shapes that could result from different object poses. We also test Tac2Pose on object models reconstructed from a 3D scanner, to evaluate the robustness to uncertainty in the object model. Finally, we demonstrate the advantages of Tac2Pose compared with three baseline methods for tactile pose estimation: directly regressing the object pose with a neural network, matching an observed contact to a set of possible contacts using a standard classification neural network, and direct pixel comparison of an observed contact with a set of possible contacts.
Website: this http URL

12.ClusterGNN: Cluster-based Coarse-to-Fine Graph Neural Network for Efficient Feature Matching ⬇️

Graph Neural Networks (GNNs) with attention have been successfully applied for learning visual feature matching. However, current methods learn with complete graphs, resulting in a quadratic complexity in the number of features. Motivated by a prior observation that self- and cross- attention matrices converge to a sparse representation, we propose ClusterGNN, an attentional GNN architecture which operates on clusters for learning the feature matching task. Using a progressive clustering module we adaptively divide keypoints into different subgraphs to reduce redundant connectivity, and employ a coarse-to-fine paradigm for mitigating miss-classification within images. Our approach yields a 59.7% reduction in runtime and 58.4% reduction in memory consumption for dense detection, compared to current state-of-the-art GNN-based matching, while achieving a competitive performance on various computer vision tasks.

13.Estimation of Reliable Proposal Quality for Temporal Action Detection ⬇️

Temporal action detection (TAD) aims to locate and recognize the actions in an untrimmed video. Anchor-free methods have made remarkable progress which mainly formulate TAD into two tasks: classification and localization using two separate branches. This paper reveals the temporal misalignment between the two tasks hindering further progress. To address this, we propose a new method that gives insights into moment and region perspectives simultaneously to align the two tasks by acquiring reliable proposal quality. For the moment perspective, Boundary Evaluate Module (BEM) is designed which focuses on local appearance and motion evolvement to estimate boundary quality and adopts a multi-scale manner to deal with varied action durations. For the region perspective, we introduce Region Evaluate Module (REM) which uses a new and efficient sampling method for proposal feature representation containing more contextual information compared with point feature to refine category score and proposal boundary. The proposed Boundary Evaluate Module and Region Evaluate Module (BREM) are generic, and they can be easily integrated with other anchor-free TAD methods to achieve superior performance. In our experiments, BREM is combined with two different frameworks and improves the performance on THUMOS14 by 3.6$%$ and 1.0$%$ respectively, reaching a new state-of-the-art (63.6$%$ average $m$AP). Meanwhile, a competitive result of 36.2% average $m$AP is achieved on ActivityNet-1.3 with the consistent improvement of BREM.

14.Multi-Head Distillation for Continual Unsupervised Domain Adaptation in Semantic Segmentation ⬇️

Unsupervised Domain Adaptation (UDA) is a transfer learning task which aims at training on an unlabeled target domain by leveraging a labeled source domain. Beyond the traditional scope of UDA with a single source domain and a single target domain, real-world perception systems face a variety of scenarios to handle, from varying lighting conditions to many cities around the world. In this context, UDAs with several domains increase the challenges with the addition of distribution shifts within the different target domains. This work focuses on a novel framework for learning UDA, continuous UDA, in which models operate on multiple target domains discovered sequentially, without access to previous target domains. We propose MuHDi, for Multi-Head Distillation, a method that solves the catastrophic forgetting problem, inherent in continual learning tasks. MuHDi performs distillation at multiple levels from the previous model as well as an auxiliary target-specialist segmentation head. We report both extensive ablation and experiments on challenging multi-target UDA semantic segmentation benchmarks to validate the proposed learning scheme and architecture.

15.Loss-based Sequential Learning for Active Domain Adaptation ⬇️

Active domain adaptation (ADA) studies have mainly addressed query selection while following existing domain adaptation strategies. However, we argue that it is critical to consider not only query selection criteria but also domain adaptation strategies designed for ADA scenarios. This paper introduces sequential learning considering both domain type (source/target) or labelness (labeled/unlabeled). We first train our model only on labeled target samples obtained by loss-based query selection. When loss-based query selection is applied under domain shift, unuseful high-loss samples gradually increase, and the labeled-sample diversity becomes low. To solve these, we fully utilize pseudo labels of the unlabeled target domain by leveraging loss prediction. We further encourage pseudo labels to have low self-entropy and diverse class distributions. Our model significantly outperforms previous methods as well as baseline models in various benchmark datasets.

16.Hybrid ISTA: Unfolding ISTA With Convergence Guarantees Using Free-Form Deep Neural Networks ⬇️

It is promising to solve linear inverse problems by unfolding iterative algorithms (e.g., iterative shrinkage thresholding algorithm (ISTA)) as deep neural networks (DNNs) with learnable parameters. However, existing ISTA-based unfolded algorithms restrict the network architectures for iterative updates with the partial weight coupling structure to guarantee convergence. In this paper, we propose hybrid ISTA to unfold ISTA with both pre-computed and learned parameters by incorporating free-form DNNs (i.e., DNNs with arbitrary feasible and reasonable network architectures), while ensuring theoretical convergence. We first develop HCISTA to improve the efficiency and flexibility of classical ISTA (with pre-computed parameters) without compromising the convergence rate in theory. Furthermore, the DNN-based hybrid algorithm is generalized to popular variants of learned ISTA, dubbed HLISTA, to enable a free architecture of learned parameters with a guarantee of linear convergence. To our best knowledge, this paper is the first to provide a convergence-provable framework that enables free-form DNNs in ISTA-based unfolded algorithms. This framework is general to endow arbitrary DNNs for solving linear inverse problems with convergence guarantees. Extensive experiments demonstrate that hybrid ISTA can reduce the reconstruction error with an improved convergence rate in the tasks of sparse recovery and compressive sensing.

17.Multi-Layer Modeling of Dense Vegetation from Aerial LiDAR Scans ⬇️

The analysis of the multi-layer structure of wild forests is an important challenge of automated large-scale forestry. While modern aerial LiDARs offer geometric information across all vegetation layers, most datasets and methods focus only on the segmentation and reconstruction of the top of canopy. We release WildForest3D, which consists of 29 study plots and over 2000 individual trees across 47 000m2 with dense 3D annotation, along with occupancy and height maps for 3 vegetation layers: ground vegetation, understory, and overstory. We propose a 3D deep network architecture predicting for the first time both 3D point-wise labels and high-resolution layer occupancy rasters simultaneously. This allows us to produce a precise estimation of the thickness of each vegetation layer as well as the corresponding watertight meshes, therefore meeting most forestry purposes. Both the dataset and the model are released in open access: this https URL.

18.A Simple Structure For Building A Robust Model ⬇️

As deep learning applications, especially programs of computer vision, are increasingly deployed in our lives, we have to think more urgently about the security of these applications.One effective way to improve the security of deep learning models is to perform adversarial training, which allows the model to be compatible with samples that are deliberately created for use in attacking the model.Based on this, we propose a simple architecture to build a model with a certain degree of robustness, which improves the robustness of the trained network by adding an adversarial sample detection network for cooperative this http URL the same time, we design a new data sampling strategy that incorporates multiple existing attacks, allowing the model to adapt to many different adversarial attacks with a single training.We conducted some experiments to test the effectiveness of this design based on Cifar10 dataset, and the results indicate that it has some degree of positive effect on the robustness of the model.Our code could be found at this https URL.

19.Unsupervised Domain Adaptation for Monocular 3D Object Detection via Self-Training ⬇️

Monocular 3D object detection (Mono3D) has achieved unprecedented success with the advent of deep learning techniques and emerging large-scale autonomous driving datasets. However, drastic performance degradation remains an unwell-studied challenge for practical cross-domain deployment as the lack of labels on the target domain. In this paper, we first comprehensively investigate the significant underlying factor of the domain gap in Mono3D, where the critical observation is a depth-shift issue caused by the geometric misalignment of domains. Then, we propose STMono3D, a new self-teaching framework for unsupervised domain adaptation on Mono3D. To mitigate the depth-shift, we introduce the geometry-aligned multi-scale training strategy to disentangle the camera parameters and guarantee the geometry consistency of domains. Based on this, we develop a teacher-student paradigm to generate adaptive pseudo labels on the target domain. Benefiting from the end-to-end framework that provides richer information of the pseudo labels, we propose the quality-aware supervision strategy to take instance-level pseudo confidences into account and improve the effectiveness of the target-domain training process. Moreover, the positive focusing training strategy and dynamic threshold are proposed to handle tremendous FN and FP pseudo samples. STMono3D achieves remarkable performance on all evaluated datasets and even surpasses fully supervised results on the KITTI 3D object detection dataset. To the best of our knowledge, this is the first study to explore effective UDA methods for Mono3D.

20.Graph-DETR3D: Rethinking Overlapping Regions for Multi-View 3D Object Detection ⬇️

3D object detection from multiple image views is a fundamental and challenging task for visual scene understanding. Due to its low cost and high efficiency, multi-view 3D object detection has demonstrated promising application prospects. However, accurately detecting objects through perspective views in the 3D space is extremely difficult due to the lack of depth information. Recently, DETR3D introduces a novel 3D-2D query paradigm in aggregating multi-view images for 3D object detection and achieves state-of-the-art performance. In this paper, with intensive pilot experiments, we quantify the objects located at different regions and find that the "truncated instances" (i.e., at the border regions of each image) are the main bottleneck hindering the performance of DETR3D. Although it merges multiple features from two adjacent views in the overlapping regions, DETR3D still suffers from insufficient feature aggregation, thus missing the chance to fully boost the detection performance. In an effort to tackle the problem, we propose Graph-DETR3D to automatically aggregate multi-view imagery information through graph structure learning (GSL). It constructs a dynamic 3D graph between each object query and 2D feature maps to enhance the object representations, especially at the border regions. Besides, Graph-DETR3D benefits from a novel depth-invariant multi-scale training strategy, which maintains the visual depth consistency by simultaneously scaling the image size and the object depth. Extensive experiments on the nuScenes dataset demonstrate the effectiveness and efficiency of our Graph-DETR3D. Notably, our best model achieves 49.5 NDS on the nuScenes test leaderboard, achieving new state-of-the-art in comparison with various published image-view 3D object detectors.

21.Joint-Modal Label Denoising for Weakly-Supervised Audio-Visual Video Parsing ⬇️

This paper focuses on the weakly-supervised audio-visual video parsing task, which aims to recognize all events belonging to each modality and localize their temporal boundaries. This task is challenging because only overall labels indicating the video events are provided for training. However, an event might be labeled but not appear in one of the modalities, which results in a modality-specific noisy label problem. Motivated by two observations that networks tend to learn clean samples first and that a labeled event would appear in at least one modality, we propose a training strategy to identify and remove modality-specific noisy labels dynamically. Specifically, we sort the losses of all instances within a mini-batch individually in each modality, then select noisy samples according to relationships between intra-modal and inter-modal losses. Besides, we also propose a simple but valid noise ratio estimation method by calculating the proportion of instances whose confidence is below a preset threshold. Our method makes large improvements over the previous state of the arts (e.g., from 60.0% to 63.8% in segment-level visual metric), which demonstrates the effectiveness of our approach.

22.Goal-driven Self-Attentive Recurrent Networks for Trajectory Prediction ⬇️

Human trajectory forecasting is a key component of autonomous vehicles, social-aware robots and advanced video-surveillance applications. This challenging task typically requires knowledge about past motion, the environment and likely destination areas. In this context, multi-modality is a fundamental aspect and its effective modeling can be beneficial to any architecture. Inferring accurate trajectories is nevertheless challenging, due to the inherently uncertain nature of the future. To overcome these difficulties, recent models use different inputs and propose to model human intentions using complex fusion mechanisms. In this respect, we propose a lightweight attention-based recurrent backbone that acts solely on past observed positions. Although this backbone already provides promising results, we demonstrate that its prediction accuracy can be improved considerably when combined with a scene-aware goal-estimation module. To this end, we employ a common goal module, based on a U-Net architecture, which additionally extracts semantic information to predict scene-compliant destinations. We conduct extensive experiments on publicly-available datasets (i.e. SDD, inD, ETH/UCY) and show that our approach performs on par with state-of-the-art techniques while reducing model complexity.

23.PedRecNet: Multi-task deep neural network for full 3D human pose and orientation estimation ⬇️

We present a multitask network that supports various deep neural network based pedestrian detection functions. Besides 2D and 3D human pose, it also supports body and head orientation estimation based on full body bounding box input. This eliminates the need for explicit face recognition. We show that the performance of 3D human pose estimation and orientation estimation is comparable to the state-of-the-art. Since very few data sets exist for 3D human pose and in particular body and head orientation estimation based on full body data, we further show the benefit of particular simulation data to train the network. The network architecture is relatively simple, yet powerful, and easily adaptable for further research and applications.

24.Rethinking Multi-Modal Alignment in Video Question Answering from Feature and Sample Perspectives ⬇️

Reasoning about causal and temporal event relations in videos is a new destination of Video Question Answering (VideoQA).The major stumbling block to achieve this purpose is the semantic gap between language and video since they are at different levels of abstraction. Existing efforts mainly focus on designing sophisticated architectures while utilizing frame- or object-level visual representations. In this paper, we reconsider the multi-modal alignment problem in VideoQA from feature and sample perspectives to achieve better performance. From the view of feature,we break down the video into trajectories and first leverage trajectory feature in VideoQA to enhance the alignment between two modalities. Moreover, we adopt a heterogeneous graph architecture and design a hierarchical framework to align both trajectory-level and frame-level visual feature with language feature. In addition, we found that VideoQA models are largely dependent on language priors and always neglect visual-language interactions. Thus, two effective yet portable training augmentation strategies are designed to strengthen the cross-modal correspondence ability of our model from the view of sample. Extensive results show that our method outperforms all the state-of-the-art models on the challenging NExT-QA benchmark, which demonstrates the effectiveness of the proposed method.

25.Combining Visual Saliency Methods and Sparse Keypoint Annotations to Providently Detect Vehicles at Night ⬇️

Provident detection of other road users at night has the potential for increasing road safety. For this purpose, humans intuitively use visual cues, such as light cones and light reflections emitted by other road users to be able to react to oncoming traffic at an early stage. This behavior can be imitated by computer vision methods by predicting the appearance of vehicles based on emitted light reflections caused by the vehicle's headlights. Since current object detection algorithms are mainly based on detecting directly visible objects annotated via bounding boxes, the detection and annotation of light reflections without sharp boundaries is challenging. For this reason, the extensive open-source dataset PVDN (Provident Vehicle Detection at Night) was published, which includes traffic scenarios at night with light reflections annotated via keypoints. In this paper, we explore the potential of saliency-based approaches to create different object representations based on the visual saliency and sparse keypoint annotations of the PVDN dataset. For that, we extend the general idea of Boolean map saliency towards a context-aware approach by taking into consideration sparse keypoint annotations by humans. We show that this approach allows for an automated derivation of different object representations, such as binary maps or bounding boxes so that detection models can be trained on different annotation variants and the problem of providently detecting vehicles at night can be tackled from different perspectives. With that, we provide further powerful tools and methods to study the problem of detecting vehicles at night before they are actually visible.

26.VITA: A Multi-Source Vicinal Transfer Augmentation Method for Out-of-Distribution Generalization ⬇️

Invariance to diverse types of image corruption, such as noise, blurring, or colour shifts, is essential to establish robust models in computer vision. Data augmentation has been the major approach in improving the robustness against common corruptions. However, the samples produced by popular augmentation strategies deviate significantly from the underlying data manifold. As a result, performance is skewed toward certain types of corruption. To address this issue, we propose a multi-source vicinal transfer augmentation (VITA) method for generating diverse on-manifold samples. The proposed VITA consists of two complementary parts: tangent transfer and integration of multi-source vicinal samples. The tangent transfer creates initial augmented samples for improving corruption robustness. The integration employs a generative model to characterize the underlying manifold built by vicinal samples, facilitating the generation of on-manifold samples. Our proposed VITA significantly outperforms the current state-of-the-art augmentation methods, demonstrated in extensive experiments on corruption benchmarks.

27.A Spatio-Temporal Multilayer Perceptron for Gesture Recognition ⬇️

Gesture recognition is essential for the interaction of autonomous vehicles with humans. While the current approaches focus on combining several modalities like image features, keypoints and bone vectors, we present neural network architecture that delivers state-of-the-art results only with body skeleton input data. We propose the spatio-temporal multilayer perceptron for gesture recognition in the context of autonomous vehicles. Given 3D body poses over time, we define temporal and spatial mixing operations to extract features in both domains. Additionally, the importance of each time step is re-weighted with Squeeze-and-Excitation layers. An extensive evaluation of the TCG and Drive&Act datasets is provided to showcase the promising performance of our approach. Furthermore, we deploy our model to our autonomous vehicle to show its real-time capability and stable execution.

28.IMDeception: Grouped Information Distilling Super-Resolution Network ⬇️

Single-Image-Super-Resolution (SISR) is a classical computer vision problem that has benefited from the recent advancements in deep learning methods, especially the advancements of convolutional neural networks (CNN). Although state-of-the-art methods improve the performance of SISR on several datasets, direct application of these networks for practical use is still an issue due to heavy computational load. For this purpose, recently, researchers have focused on more efficient and high-performing network structures. Information multi-distilling network (IMDN) is one of the highly efficient SISR networks with high performance and low computational load. IMDN achieves this efficiency with various mechanisms such as Intermediate Information Collection (IIC), working in a global setting, Progressive Refinement Module (PRM), and Contrast Aware Channel Attention (CCA), employed in a local setting. These mechanisms, however, do not equally contribute to the efficiency and performance of IMDN. In this work, we propose the Global Progressive Refinement Module (GPRM) as a less parameter-demanding alternative to the IIC module for feature aggregation. To further decrease the number of parameters and floating point operations persecond (FLOPS), we also propose Grouped Information Distilling Blocks (GIDB). Using the proposed structures, we design an efficient SISR network called IMDeception. Experiments reveal that the proposed network performs on par with state-of-the-art models despite having a limited number of parameters and FLOPS. Furthermore, using grouped convolutions as a building block of GIDB increases room for further optimization during deployment. To show its potential, the proposed model was deployed on NVIDIA Jetson Xavier AGX and it has been shown that it can run in real-time on this edge device

29.OCFormer: One-Class Transformer Network for Image Classification ⬇️

We propose a novel deep learning framework based on Vision Transformers (ViT) for one-class classification. The core idea is to use zero-centered Gaussian noise as a pseudo-negative class for latent space representation and then train the network using the optimal loss function. In prior works, there have been tremendous efforts to learn a good representation using varieties of loss functions, which ensures both discriminative and compact properties. The proposed one-class Vision Transformer (OCFormer) is exhaustively experimented on CIFAR-10, CIFAR-100, Fashion-MNIST and CelebA eyeglasses datasets. Our method has shown significant improvements over competing CNN based one-class classifier approaches.

30.Fine-tuning Pruned Networks with Linear Over-parameterization ⬇️

Structured pruning compresses neural networks by reducing channels (filters) for fast inference and low footprint at run-time. To restore accuracy after pruning, fine-tuning is usually applied to pruned networks. However, too few remaining parameters in pruned networks inevitably bring a great challenge to fine-tuning to restore accuracy. To address this challenge, we propose a novel method that first linearly over-parameterizes the compact layers in pruned networks to enlarge the number of fine-tuning parameters and then re-parameterizes them to the original layers after fine-tuning. Specifically, we equivalently expand the convolution/linear layer with several consecutive convolution/linear layers that do not alter the current output feature maps. Furthermore, we utilize similarity-preserving knowledge distillation that encourages the over-parameterized block to learn the immediate data-to-data similarities of the corresponding dense layer to maintain its feature learning ability. The proposed method is comprehensively evaluated on CIFAR-10 and ImageNet which significantly outperforms the vanilla fine-tuning strategy, especially for large pruning ratio.

31.SwinFuse: A Residual Swin Transformer Fusion Network for Infrared and Visible Images ⬇️

The existing deep learning fusion methods mainly concentrate on the convolutional neural networks, and few attempts are made with transformer. Meanwhile, the convolutional operation is a content-independent interaction between the image and convolution kernel, which may lose some important contexts and further limit fusion performance. Towards this end, we present a simple and strong fusion baseline for infrared and visible images, namely\textit{ Residual Swin Transformer Fusion Network}, termed as SwinFuse. Our SwinFuse includes three parts: the global feature extraction, fusion layer and feature reconstruction. In particular, we build a fully attentional feature encoding backbone to model the long-range dependency, which is a pure transformer network and has a stronger representation ability compared with the convolutional neural networks. Moreover, we design a novel feature fusion strategy based on $L_{1}$-norm for sequence matrices, and measure the corresponding activity levels from row and column vector dimensions, which can well retain competitive infrared brightness and distinct visible details. Finally, we testify our SwinFuse with nine state-of-the-art traditional and deep learning methods on three different datasets through subjective observations and objective comparisons, and the experimental results manifest that the proposed SwinFuse obtains surprising fusion performance with strong generalization ability and competitive computational efficiency. The code will be available at this https URL.

32.Joint Feature Distribution Alignment Learning for NIR-VIS and VIS-VIS Face Recognition ⬇️

Face recognition for visible light (VIS) images achieve high accuracy thanks to the recent development of deep learning. However, heterogeneous face recognition (HFR), which is a face matching in different domains, is still a difficult task due to the domain discrepancy and lack of large HFR dataset. Several methods have attempted to reduce the domain discrepancy by means of fine-tuning, which causes significant degradation of the performance in the VIS domain because it loses the highly discriminative VIS representation. To overcome this problem, we propose joint feature distribution alignment learning (JFDAL) which is a joint learning approach utilizing knowledge distillation. It enables us to achieve high HFR performance with retaining the original performance for the VIS domain. Extensive experiments demonstrate that our proposed method delivers statistically significantly better performances compared with the conventional fine-tuning approach on a public HFR dataset Oulu-CASIA NIR&VIS and popular verification datasets in VIS domain such as FLW, CFP, AgeDB. Furthermore, comparative experiments with existing state-of-the-art HFR methods show that our method achieves a comparable HFR performance on the Oulu-CASIA NIR&VIS dataset with less degradation of VIS performance.

33.Surpassing the Human Accuracy: Detecting Gallbladder Cancer from USG Images with Curriculum Learning ⬇️

We explore the potential of CNN-based models for gallbladder cancer (GBC) detection from ultrasound (USG) images as no prior study is known. USG is the most common diagnostic modality for GB diseases due to its low cost and accessibility. However, USG images are challenging to analyze due to low image quality, noise, and varying viewpoints due to the handheld nature of the sensor. Our exhaustive study of state-of-the-art (SOTA) image classification techniques for the problem reveals that they often fail to learn the salient GB region due to the presence of shadows in the USG images. SOTA object detection techniques also achieve low accuracy because of spurious textures due to noise or adjacent organs. We propose GBCNet to tackle the challenges in our problem. GBCNet first extracts the regions of interest (ROIs) by detecting the GB (and not the cancer), and then uses a new multi-scale, second-order pooling architecture specializing in classifying GBC. To effectively handle spurious textures, we propose a curriculum inspired by human visual acuity, which reduces the texture biases in GBCNet. Experimental results demonstrate that GBCNet significantly outperforms SOTA CNN models, as well as the expert radiologists. Our technical innovations are generic to other USG image analysis tasks as well. Hence, as a validation, we also show the efficacy of GBCNet in detecting breast cancer from USG images. Project page with source code, trained models, and data is available at this https URL

34.Unsupervised Hierarchical Semantic Segmentation with Multiview Cosegmentation and Clustering Transformers ⬇️

Unsupervised semantic segmentation aims to discover groupings within and across images that capture object and view-invariance of a category without external supervision. Grouping naturally has levels of granularity, creating ambiguity in unsupervised segmentation. Existing methods avoid this ambiguity and treat it as a factor outside modeling, whereas we embrace it and desire hierarchical grouping consistency for unsupervised segmentation.
We approach unsupervised segmentation as a pixel-wise feature learning problem. Our idea is that a good representation shall reveal not just a particular level of grouping, but any level of grouping in a consistent and predictable manner. We enforce spatial consistency of grouping and bootstrap feature learning with co-segmentation among multiple views of the same image, and enforce semantic consistency across the grouping hierarchy with clustering transformers between coarse- and fine-grained features.
We deliver the first data-driven unsupervised hierarchical semantic segmentation method called Hierarchical Segment Grouping (HSG). Capturing visual similarity and statistical co-occurrences, HSG also outperforms existing unsupervised segmentation methods by a large margin on five major object- and scene-centric benchmarks. Our code is publicly available at this https URL .

35.Audio-Visual Scene Classification Using A Transfer Learning Based Joint Optimization Strategy ⬇️

Recently, audio-visual scene classification (AVSC) has attracted increasing attention from multidisciplinary communities. Previous studies tended to adopt a pipeline training strategy, which uses well-trained visual and acoustic encoders to extract high-level representations (embeddings) first, then utilizes them to train the audio-visual classifier. In this way, the extracted embeddings are well suited for uni-modal classifiers, but not necessarily suited for multi-modal ones. In this paper, we propose a joint training framework, using the acoustic features and raw images directly as inputs for the AVSC task. Specifically, we retrieve the bottom layers of pre-trained image models as visual encoder, and jointly optimize the scene classifier and 1D-CNN based acoustic encoder during training. We evaluate the approach on the development dataset of TAU Urban Audio-Visual Scenes 2021. The experimental results show that our proposed approach achieves significant improvement over the conventional pipeline training strategy. Moreover, our best single system outperforms previous state-of-the-art methods, yielding a log loss of 0.1517 and accuracy of 94.59% on the official test fold.

36.Single Object Tracking Research: A Survey ⬇️

Visual object tracking is an important task in computer vision, which has many real-world applications, e.g., video surveillance, visual navigation. Visual object tracking also has many challenges, e.g., object occlusion and deformation. To solve above problems and track the target accurately and efficiently, many tracking algorithms have emerged in recent years. This paper presents the rationale and representative works of two most popular tracking frameworks in past ten years, i.e., the corelation filter and Siamese network for object tracking. Then we present some deep learning based tracking methods categorized by different network structures. We also introduce some classical strategies for handling the challenges in tracking problem. Further, this paper detailedly present and compare the benchmarks and challenges for tracking, from which we summarize the development history and development trend of visual tracking. Focusing on the future development of object tracking, which we think would be applied in real-world scenes before some problems to be addressed, such as the problems in long-term tracking, low-power high-speed tracking and attack-robust tracking. In the future, the integration of multimodal data, e.g., the depth image, thermal image with traditional color image, will provide more solutions for visual tracking. Moreover, tracking task will go together with some other tasks, e.g., video object detection and segmentation.

37.Dynamic Point Cloud Compression with Cross-Sectional Approach ⬇️

The recent development of dynamic point clouds has introduced the possibility of mimicking natural reality, and greatly assisting quality of life. However, to broadcast successfully, the dynamic point clouds require higher compression due to their huge volume of data compared to the traditional video. Recently, MPEG finalized a Video-based Point Cloud Compression standard known as V-PCC. However, V-PCC requires huge computational time due to expensive normal calculation and segmentation, sacrifices some points to limit the number of 2D patches, and cannot occupy all spaces in the 2D frame. The proposed method addresses these limitations by using a novel cross-sectional approach. This approach reduces expensive normal estimation and segmentation, retains more points, and utilizes more spaces for 2D frame generation compared to the VPCC. The experimental results using standard video sequences show that the proposed technique can achieve better compression in both geometric and texture data compared to the V-PCC standard.

38.PointInst3D: Segmenting 3D Instances by Points ⬇️

The current state-of-the-art methods in 3D instance segmentation typically involve a clustering step, despite the tendency towards heuristics, greedy algorithms, and a lack of robustness to the changes in data statistics. In contrast, we propose a fully-convolutional 3D point cloud instance segmentation method that works in a per-point prediction fashion. In doing so it avoids the challenges that clustering-based methods face: introducing dependencies among different tasks of the model. We find the key to its success is assigning a suitable target to each sampled point. Instead of the commonly used static or distance-based assignment strategies, we propose to use an Optimal Transport approach to optimally assign target masks to the sampled points according to the dynamic matching costs. Our approach achieves promising results on both ScanNet and S3DIS benchmarks. The proposed approach removes intertask dependencies and thus represents a simpler and more flexible 3D instance segmentation framework than other competing methods, while achieving improved segmentation accuracy.

39.Video Frame Interpolation Based on Deformable Kernel Region ⬇️

Video frame interpolation task has recently become more and more prevalent in the computer vision field. At present, a number of researches based on deep learning have achieved great success. Most of them are either based on optical flow information, or interpolation kernel, or a combination of these two methods. However, these methods have ignored that there are grid restrictions on the position of kernel region during synthesizing each target pixel. These limitations result in that they cannot well adapt to the irregularity of object shape and uncertainty of motion, which may lead to irrelevant reference pixels used for interpolation. In order to solve this problem, we revisit the deformable convolution for video interpolation, which can break the fixed grid restrictions on the kernel region, making the distribution of reference points more suitable for the shape of the object, and thus warp a more accurate interpolation frame. Experiments are conducted on four datasets to demonstrate the superior performance of the proposed model in comparison to the state-of-the-art alternatives.

40.DRT: A Lightweight Single Image Deraining Recursive Transformer ⬇️

Over parameterization is a common technique in deep learning to help models learn and generalize sufficiently to the given task; nonetheless, this often leads to enormous network structures and consumes considerable computing resources during training. Recent powerful transformer-based deep learning models on vision tasks usually have heavy parameters and bear training difficulty. However, many dense-prediction low-level computer vision tasks, such as rain streak removing, often need to be executed on devices with limited computing power and memory in practice. Hence, we introduce a recursive local window-based self-attention structure with residual connections and propose deraining a recursive transformer (DRT), which enjoys the superiority of the transformer but requires a small amount of computing resources. In particular, through recursive architecture, our proposed model uses only 1.3% of the number of parameters of the current best performing model in deraining while exceeding the state-of-the-art methods on the Rain100L benchmark by at least 0.33 dB. Ablation studies also investigate the impact of recursions on derain outcomes. Moreover, since the model contains no deliberate design for deraining, it can also be applied to other image restoration tasks. Our experiment shows that it can achieve competitive results on desnowing. The source code and pretrained model can be found at this https URL.

41.Deep Reinforcement Learning Using a Low-Dimensional Observation Filter for Visual Complex Video Game Playing ⬇️

Deep Reinforcement Learning (DRL) has produced great achievements since it was proposed, including the possibility of processing raw vision input data. However, training an agent to perform tasks based on image feedback remains a challenge. It requires the processing of large amounts of data from high-dimensional observation spaces, frame by frame, and the agent's actions are computed according to deep neural network policies, end-to-end. Image pre-processing is an effective way of reducing these high dimensional spaces, eliminating unnecessary information present in the scene, supporting the extraction of features and their representations in the agent's neural network. Modern video-games are examples of this type of challenge for DRL algorithms because of their visual complexity. In this paper, we propose a low-dimensional observation filter that allows a deep Q-network agent to successfully play in a visually complex and modern video-game, called Neon Drive.

42.Simulating Fluids in Real-World Still Images ⬇️

In this work, we tackle the problem of real-world fluid animation from a still image. The key of our system is a surface-based layered representation deriving from video decomposition, where the scene is decoupled into a surface fluid layer and an impervious background layer with corresponding transparencies to characterize the composition of the two layers. The animated video can be produced by warping only the surface fluid layer according to the estimation of fluid motions and recombining it with the background. In addition, we introduce surface-only fluid simulation, a $2.5D$ fluid calculation version, as a replacement for motion estimation. Specifically, we leverage the triangular mesh based on a monocular depth estimator to represent the fluid surface layer and simulate the motion in the physics-based framework with the inspiration of the classic theory of the hybrid Lagrangian-Eulerian method, along with a learnable network so as to adapt to complex real-world image textures. We demonstrate the effectiveness of the proposed system through comparison with existing methods in both standard objective metrics and subjective ranking scores. Extensive experiments not only indicate our method's competitive performance for common fluid scenes but also better robustness and reasonability under complex transparent fluid scenarios. Moreover, as the proposed surface-based layer representation and surface-only fluid simulation naturally disentangle the scene, interactive editing such as adding objects to the river and texture replacing could be easily achieved with realistic results.

43.EMOCA: Emotion Driven Monocular Face Capture and Animation ⬇️

As 3D facial avatars become more widely used for communication, it is critical that they faithfully convey emotion. Unfortunately, the best recent methods that regress parametric 3D face models from monocular images are unable to capture the full spectrum of facial expression, such as subtle or extreme emotions. We find the standard reconstruction metrics used for training (landmark reprojection error, photometric error, and face recognition loss) are insufficient to capture high-fidelity expressions. The result is facial geometries that do not match the emotional content of the input image. We address this with EMOCA (EMOtion Capture and Animation), by introducing a novel deep perceptual emotion consistency loss during training, which helps ensure that the reconstructed 3D expression matches the expression depicted in the input image. While EMOCA achieves 3D reconstruction errors that are on par with the current best methods, it significantly outperforms them in terms of the quality of the reconstructed expression and the perceived emotional content. We also directly regress levels of valence and arousal and classify basic expressions from the estimated 3D face parameters. On the task of in-the-wild emotion recognition, our purely geometric approach is on par with the best image-based methods, highlighting the value of 3D geometry in analyzing human behavior. The model and code are publicly available at this https URL.

44.Large Scale Time-Series Representation Learning via Simultaneous Low and High Frequency Feature Bootstrapping ⬇️

Learning representation from unlabeled time series data is a challenging problem. Most existing self-supervised and unsupervised approaches in the time-series domain do not capture low and high-frequency features at the same time. Further, some of these methods employ large scale models like transformers or rely on computationally expensive techniques such as contrastive learning. To tackle these problems, we propose a non-contrastive self-supervised learning approach efficiently captures low and high-frequency time-varying features in a cost-effective manner. Our method takes raw time series data as input and creates two different augmented views for two branches of the model, by randomly sampling the augmentations from same family. Following the terminology of BYOL, the two branches are called online and target network which allows bootstrapping of the latent representation. In contrast to BYOL, where a backbone encoder is followed by multilayer perceptron (MLP) heads, the proposed model contains additional temporal convolutional network (TCN) heads. As the augmented views are passed through large kernel convolution blocks of the encoder, the subsequent combination of MLP and TCN enables an effective representation of low as well as high-frequency time-varying features due to the varying receptive fields. The two modules (MLP and TCN) act in a complementary manner. We train an online network where each module learns to predict the outcome of the respective module of target network branch. To demonstrate the robustness of our model we performed extensive experiments and ablation studies on five real-world time-series datasets. Our method achieved state-of-art performance on all five real-world datasets.

45.Towards the Semantic Weak Generalization Problem in Generative Zero-Shot Learning: Ante-hoc and Post-hoc ⬇️

In this paper, we present a simple and effective strategy lowering the previously unexplored factors that limit the performance ceiling of generative Zero-Shot Learning (ZSL). We begin by formally defining semantic generalization, then look into approaches for reducing the semantic weak generalization problem and minimizing its negative influence on classifier training. In the ante-hoc phase, we augment the generator's semantic input, as well as relax the fitting target of the generator. In the post-hoc phase (after generating simulated unseen samples), we derive from the gradient of the loss function to minimize the gradient increment on seen classifier weights carried by biased unseen distribution, which tends to cause misleading on intra-seen class decision boundaries. Without complicated designs, our approach hit the essential problem and significantly outperform the state-of-the-art on four widely used ZSL datasets.

46.RMGN: A Regional Mask Guided Network for Parser-free Virtual Try-on ⬇️

Virtual try-on(VTON) aims at fitting target clothes to reference person images, which is widely adopted in e-commerce.Existing VTON approaches can be narrowly categorized into Parser-Based(PB) and Parser-Free(PF) by whether relying on the parser information to mask the persons' clothes and synthesize try-on images. Although abandoning parser information has improved the applicability of PF methods, the ability of detail synthesizing has also been sacrificed. As a result, the distraction from original cloth may persistin synthesized images, especially in complicated postures and high resolution applications. To address the aforementioned issue, we propose a novel PF method named Regional Mask Guided Network(RMGN). More specifically, a regional mask is proposed to explicitly fuse the features of target clothes and reference persons so that the persisted distraction can be eliminated. A posture awareness loss and a multi-level feature extractor are further proposed to handle the complicated postures and synthesize high resolution images. Extensive experiments demonstrate that our proposed RMGN outperforms both state-of-the-art PB and PF methods.Ablation studies further verify the effectiveness ofmodules in RMGN.

47.Source-Free Domain Adaptation via Distribution Estimation ⬇️

Domain Adaptation aims to transfer the knowledge learned from a labeled source domain to an unlabeled target domain whose data distributions are different. However, the training data in source domain required by most of the existing methods is usually unavailable in real-world applications due to privacy preserving policies. Recently, Source-Free Domain Adaptation (SFDA) has drawn much attention, which tries to tackle domain adaptation problem without using source data. In this work, we propose a novel framework called SFDA-DE to address SFDA task via source Distribution Estimation. Firstly, we produce robust pseudo-labels for target data with spherical k-means clustering, whose initial class centers are the weight vectors (anchors) learned by the classifier of pretrained model. Furthermore, we propose to estimate the class-conditioned feature distribution of source domain by exploiting target data and corresponding anchors. Finally, we sample surrogate features from the estimated distribution, which are then utilized to align two domains by minimizing a contrastive adaptation loss function. Extensive experiments show that the proposed method achieves state-of-the-art performance on multiple DA benchmarks, and even outperforms traditional DA methods which require plenty of source data.

48.Lesion Localization in OCT by Semi-Supervised Object Detection ⬇️

Over 300 million people worldwide are affected by various retinal diseases. By noninvasive Optical Coherence Tomography (OCT) scans, a number of abnormal structural changes in the retina, namely retinal lesions, can be identified. Automated lesion localization in OCT is thus important for detecting retinal diseases at their early stage. To conquer the lack of manual annotation for deep supervised learning, this paper presents a first study on utilizing semi-supervised object detection (SSOD) for lesion localization in OCT images. To that end, we develop a taxonomy to provide a unified and structured viewpoint of the current SSOD methods, and consequently identify key modules in these methods. To evaluate the influence of these modules in the new task, we build OCT-SS, a new dataset consisting of over 1k expert-labeled OCT B-scan images and over 13k unlabeled B-scans. Extensive experiments on OCT-SS identify Unbiased Teacher (UnT) as the best current SSOD method for lesion localization. Moreover, we improve over this strong baseline, with mAP increased from 49.34 to 50.86.

49.RealNet: Combining Optimized Object Detection with Information Fusion Depth Estimation Co-Design Method on IoT ⬇️

Depth Estimation and Object Detection Recognition play an important role in autonomous driving technology under the guidance of deep learning artificial intelligence. We propose a hybrid structure called RealNet: a co-design method combining the model-streamlined recognition algorithm, the depth estimation algorithm with information fusion, and deploying them on the Jetson-Nano for unmanned vehicles with monocular vision sensors. We use ROS for experiment. The method proposed in this paper is suitable for mobile platforms with high real-time request. Innovation of our method is using information fusion to compensate the problem of insufficient frame rate of output image, and improve the robustness of target detection and depth estimation under monocular vision.Object Detection is based on YOLO-v5. We have simplified the network structure of its DarkNet53 and realized a prediction speed up to 0.01s. Depth Estimation is based on the VNL Depth Estimation, which considers multiple geometric constraints in 3D global space. It calculates the loss function by calculating the deviation of the virtual normal vector VN and the label, which can obtain deeper depth information. We use PnP fusion algorithm to solve the problem of insufficient frame rate of depth map output. It solves the motion estimation depth from three-dimensional target to two-dimensional point based on corner feature matching, which is faster than VNL calculation. We interpolate VNL output and PnP output to achieve information fusion. Experiments show that this can effectively eliminate the jitter of depth information and improve robustness. At the control end, this method combines the results of target detection and depth estimation to calculate the target position, and uses a pure tracking control algorithm to track it.

50.Progressive Learning for Image Retrieval with Hybrid-Modality Queries ⬇️

Image retrieval with hybrid-modality queries, also known as composing text and image for image retrieval (CTI-IR), is a retrieval task where the search intention is expressed in a more complex query format, involving both vision and text modalities. For example, a target product image is searched using a reference product image along with text about changing certain attributes of the reference image as the query. It is a more challenging image retrieval task that requires both semantic space learning and cross-modal fusion. Previous approaches that attempt to deal with both aspects achieve unsatisfactory performance. In this paper, we decompose the CTI-IR task into a three-stage learning problem to progressively learn the complex knowledge for image retrieval with hybrid-modality queries. We first leverage the semantic embedding space for open-domain image-text retrieval, and then transfer the learned knowledge to the fashion-domain with fashion-related pre-training tasks. Finally, we enhance the pre-trained model from single-query to hybrid-modality query for the CTI-IR task. Furthermore, as the contribution of individual modality in the hybrid-modality query varies for different retrieval scenarios, we propose a self-supervised adaptive weighting strategy to dynamically determine the importance of image and text in the hybrid-modality query for better retrieval. Extensive experiments show that our proposed model significantly outperforms state-of-the-art methods in the mean of Recall@K by 24.9% and 9.5% on the Fashion-IQ and Shoes benchmark datasets respectively.

51.2D LiDAR and Camera Fusion Using Motion Cues for Indoor Layout Estimation ⬇️

This paper presents a novel indoor layout estimation system based on the fusion of 2D LiDAR and intensity camera data. A ground robot explores an indoor space with a single floor and vertical walls, and collects a sequence of intensity images and 2D LiDAR datasets. The LiDAR provides accurate depth information, while the camera captures high-resolution data for semantic interpretation. The alignment of sensor outputs and image segmentation are computed jointly by aligning LiDAR points, as samples of the room contour, to ground-wall boundaries in the images. The alignment problem is decoupled into a top-down view projection and a 2D similarity transformation estimation, which can be solved according to the vertical vanishing point and motion of two sensors. The recursive random sample consensus algorithm is implemented to generate, evaluate and optimize multiple hypotheses with the sequential measurements. The system allows jointly analyzing the geometric interpretation from different sensors without offline calibration. The ambiguity in images for ground-wall boundary extraction is removed with the assistance of LiDAR observations, which improves the accuracy of semantic segmentation. The localization and mapping is refined using the fused data, which enables the system to work reliably in scenes with low texture or low geometric features.

52.MVP-Human Dataset for 3D Human Avatar Reconstruction from Unconstrained Frames ⬇️

In this paper, we consider a novel problem of reconstructing a 3D human avatar from multiple unconstrained frames, independent of assumptions on camera calibration, capture space, and constrained actions. The problem should be addressed by a framework that takes multiple unconstrained images as inputs, and generates a shape-with-skinning avatar in the canonical space, finished in one feed-forward pass. To this end, we present 3D Avatar Reconstruction in the wild (ARwild), which first reconstructs the implicit skinning fields in a multi-level manner, by which the image features from multiple images are aligned and integrated to estimate a pixel-aligned implicit function that represents the clothed shape. To enable the training and testing of the new framework, we contribute a large-scale dataset, MVP-Human (Multi-View and multi-Pose 3D Human), which contains 400 subjects, each of which has 15 scans in different poses and 8-view images for each pose, providing 6,000 3D scans and 48,000 images in total. Overall, benefits from the specific network architecture and the diverse data, the trained model enables 3D avatar reconstruction from unconstrained frames and achieves state-of-the-art performance.

53.RelViT: Concept-guided Vision Transformer for Visual Relational Reasoning ⬇️

Reasoning about visual relationships is central to how humans interpret the visual world. This task remains challenging for current deep learning algorithms since it requires addressing three key technical problems jointly: 1) identifying object entities and their properties, 2) inferring semantic relations between pairs of entities, and 3) generalizing to novel object-relation combinations, i.e., systematic generalization. In this work, we use vision transformers (ViTs) as our base model for visual reasoning and make better use of concepts defined as object entities and their relations to improve the reasoning ability of ViTs. Specifically, we introduce a novel concept-feature dictionary to allow flexible image feature retrieval at training time with concept keys. This dictionary enables two new concept-guided auxiliary tasks: 1) a global task for promoting relational reasoning, and 2) a local task for facilitating semantic object-centric correspondence learning. To examine the systematic generalization of visual reasoning models, we introduce systematic splits for the standard HICO and GQA benchmarks. We show the resulting model, Concept-guided Vision Transformer (or RelViT for short) significantly outperforms prior approaches on HICO and GQA by 16% and 13% in the original split, and by 43% and 18% in the systematic split. Our ablation analyses also reveal our model's compatibility with multiple ViT variants and robustness to hyper-parameters.

54.A Survey on Unsupervised Industrial Anomaly Detection Algorithms ⬇️

Anomaly defect detection has become an indispensable part of industrial production process. In previous study, a large part of the traditional anomaly detection algorithms belong to the category of supervised learning, while the unsupervised situation is more common for most practical application scenarios. Hence gradually unsupervised anomaly detection has been the subject of much research over the last few years. In this survey, we provide a comprehensive introduction to newly proposed approaches for visual anomaly detection. We hope that it can help the research community as well as the industry field to build a broader and cross-domain perspective.

55.Supplementing Missing Visions via Dialog for Scene Graph Generations ⬇️

Most current AI systems rely on the premise that the input visual data are sufficient to achieve competitive performance in various computer vision tasks. However, the classic task setup rarely considers the challenging, yet common practical situations where the complete visual data may be inaccessible due to various reasons (e.g., restricted view range and occlusions). To this end, we investigate a computer vision task setting with incomplete visual input data. Specifically, we exploit the Scene Graph Generation (SGG) task with various levels of visual data missingness as input. While insufficient visual input intuitively leads to performance drop, we propose to supplement the missing visions via the natural language dialog interactions to better accomplish the task objective. We design a model-agnostic Supplementary Interactive Dialog (SI-Dial) framework that can be jointly learned with most existing models, endowing the current AI systems with the ability of question-answer interactions in natural language. We demonstrate the feasibility of such a task setting with missing visual input and the effectiveness of our proposed dialog module as the supplementary information source through extensive experiments and analysis, by achieving promising performance improvement over multiple baselines.

56.Can domain adaptation make object recognition work for everyone? ⬇️

Despite the rapid progress in deep visual recognition, modern computer vision datasets significantly overrepresent the developed world and models trained on such datasets underperform on images from unseen geographies. We investigate the effectiveness of unsupervised domain adaptation (UDA) of such models across geographies at closing this performance gap. To do so, we first curate two shifts from existing datasets to study the Geographical DA problem, and discover new challenges beyond data distribution shift: context shift, wherein object surroundings may change significantly across geographies, and subpopulation shift, wherein the intra-category distributions may shift. We demonstrate the inefficacy of standard DA methods at Geographical DA, highlighting the need for specialized geographical adaptation solutions to address the challenge of making object recognition work for everyone.

57.Learning Shape Priors by Pairwise Comparison for Robust Semantic Segmentation ⬇️

Semantic segmentation is important in medical image analysis. Inspired by the strong ability of traditional image analysis techniques in capturing shape priors and inter-subject similarity, many deep learning (DL) models have been recently proposed to exploit such prior information and achieved robust performance. However, these two types of important prior information are usually studied separately in existing models. In this paper, we propose a novel DL model to model both type of priors within a single framework. Specifically, we introduce an extra encoder into the classic encoder-decoder structure to form a Siamese structure for the encoders, where one of them takes a target image as input (the image-encoder), and the other concatenates a template image and its foreground regions as input (the template-encoder). The template-encoder encodes the shape priors and appearance characteristics of each foreground class in the template image. A cosine similarity based attention module is proposed to fuse the information from both encoders, to utilize both types of prior information encoded by the template-encoder and model the inter-subject similarity for each foreground class. Extensive experiments on two public datasets demonstrate that our proposed method can produce superior performance to competing methods.

58.MLP-Hash: Protecting Face Templates via Hashing of Randomized Multi-Layer Perceptron ⬇️

Applications of face recognition systems for authentication purposes are growing rapidly. Although state-of-the-art (SOTA) face recognition systems have high recognition performance, the features which are extracted for each user and are stored in the system's database contain privacy-sensitive information. Accordingly, compromising this data would jeopardize users' privacy. In this paper, we propose a new cancelable template protection method, dubbed MLP-hash, which generates protected templates by passing the extracted features through a user-specific randomly-weighted multi-layer perceptron (MLP) and binarizing the MLP output. We evaluated the unlinkability, irreversibility, and recognition performance of our proposed biometric template protection method to fulfill the ISO/IEC 30136 standard requirements. Our experiments with SOTA face recognition systems on the MOBIO and LFW datasets show that our method has competitive performance with the BioHashing and IoM Hashing (IoM-GRP and IoM-URP) template protection algorithms. We provide an open-source implementation of all the experiments presented in this paper so that other researchers can verify our findings and build upon our work.

59.Uncertain Label Correction via Auxiliary Action Unit Graphs for Facial Expression Recognition ⬇️

High-quality annotated images are significant to deep facial expression recognition (FER) methods. However, uncertain labels, mostly existing in large-scale public datasets, often mislead the training process. In this paper, we achieve uncertain label correction of facial expressions using auxiliary action unit (AU) graphs, called ULC-AG. Specifically, a weighted regularization module is introduced to highlight valid samples and suppress category imbalance in every batch. Based on the latent dependency between emotions and AUs, an auxiliary branch using graph convolutional layers is added to extract the semantic information from graph topologies. Finally, a re-labeling strategy corrects the ambiguous annotations by comparing their feature similarities with semantic templates. Experiments show that our ULC-AG achieves 89.31% and 61.57% accuracy on RAF-DB and AffectNet datasets, respectively, outperforming the baseline and state-of-the-art methods.

60.Investigating Neural Architectures by Synthetic Dataset Design ⬇️

Recent years have seen the emergence of many new neural network structures (architectures and layers). To solve a given task, a network requires a certain set of abilities reflected in its structure. The required abilities depend on each task. There is so far no systematic study of the real capacities of the proposed neural structures. The question of what each structure can and cannot achieve is only partially answered by its performance on common benchmarks. Indeed, natural data contain complex unknown statistical cues. It is therefore impossible to know what cues a given neural structure is taking advantage of in such data. In this work, we sketch a methodology to measure the effect of each structure on a network's ability, by designing ad hoc synthetic datasets. Each dataset is tailored to assess a given ability and is reduced to its simplest form: each input contains exactly the amount of information needed to solve the task. We illustrate our methodology by building three datasets to evaluate each of the three following network properties: a) the ability to link local cues to distant inferences, b) the translation covariance and c) the ability to group pixels with the same characteristics and share information among them. Using a first simplified depth estimation dataset, we pinpoint a serious nonlocal deficit of the U-Net. We then evaluate how to resolve this limitation by embedding its structure with nonlocal layers, which allow computing complex features with long-range dependencies. Using a second dataset, we compare different positional encoding methods and use the results to further improve the U-Net on the depth estimation task. The third introduced dataset serves to demonstrate the need for self-attention-like mechanisms for resolving more realistic depth estimation tasks.

61.Learning by Erasing: Conditional Entropy based Transferable Out-Of-Distribution Detection ⬇️

Out-of-distribution (OOD) detection is essential to handle the distribution shifts between training and test scenarios. For a new in-distribution (ID) dataset, existing methods require retraining to capture the dataset-specific feature representation or data distribution. In this paper, we propose a deep generative models (DGM) based transferable OOD detection method, which is unnecessary to retrain on a new ID dataset. We design an image erasing strategy to equip exclusive conditional entropy distribution for each ID dataset, which determines the discrepancy of DGM's posteriori ucertainty distribution on different ID datasets. Owing to the powerful representation capacity of convolutional neural networks, the proposed model trained on complex dataset can capture the above discrepancy between ID datasets without retraining and thus achieve transferable OOD detection. We validate the proposed method on five datasets and verity that ours achieves comparable performance to the state-of-the-art group based OOD detection methods that need to be retrained to deploy on new ID datasets. Our code is available at this https URL.

62.VISTA: Vision Transformer enhanced by U-Net and Image Colorfulness Frame Filtration for Automatic Retail Checkout ⬇️

Multi-class product counting and recognition identifies product items from images or videos for automated retail checkout. The task is challenging due to the real-world scenario of occlusions where product items overlap, fast movement in the conveyor belt, large similarity in overall appearance of the items being scanned, novel products, and the negative impact of misidentifying items. Further, there is a domain bias between training and test sets, specifically, the provided training dataset consists of synthetic images and the test set videos consist of foreign objects such as hands and tray. To address these aforementioned issues, we propose to segment and classify individual frames from a video sequence. The segmentation method consists of a unified single product item- and hand-segmentation followed by entropy masking to address the domain bias problem. The multi-class classification method is based on Vision Transformers (ViT). To identify the frames with target objects, we utilize several image processing methods and propose a custom metric to discard frames not having any product items. Combining all these mechanisms, our best system achieves 3rd place in the AI City Challenge 2022 Track 4 with an F1 score of 0.4545. Code will be available at

63.Exploring Negatives in Contrastive Learning for Unpaired Image-to-Image Translation ⬇️

Unpaired image-to-image translation aims to find a mapping between the source domain and the target domain. To alleviate the problem of the lack of supervised labels for the source images, cycle-consistency based methods have been proposed for image structure preservation by assuming a reversible relationship between unpaired images. However, this assumption only uses limited correspondence between image pairs. Recently, contrastive learning (CL) has been used to further investigate the image correspondence in unpaired image translation by using patch-based positive/negative learning. Patch-based contrastive routines obtain the positives by self-similarity computation and recognize the rest patches as negatives. This flexible learning paradigm obtains auxiliary contextualized information at a low cost. As the negatives own an impressive sample number, with curiosity, we make an investigation based on a question: are all negatives necessary for feature contrastive learning? Unlike previous CL approaches that use negatives as much as possible, in this paper, we study the negatives from an information-theoretic perspective and introduce a new negative Pruning technology for Unpaired image-to-image Translation (PUT) by sparsifying and ranking the patches. The proposed algorithm is efficient, flexible and enables the model to learn essential information between corresponding patches stably. By putting quality over quantity, only a few negative patches are required to achieve better results. Lastly, we validate the superiority, stability, and versatility of our model through comparative experiments.

64.Surface Reconstruction from Point Clouds by Learning Predictive Context Priors ⬇️

Surface reconstruction from point clouds is vital for 3D computer vision. State-of-the-art methods leverage large datasets to first learn local context priors that are represented as neural network-based signed distance functions (SDFs) with some parameters encoding the local contexts. To reconstruct a surface at a specific query location at inference time, these methods then match the local reconstruction target by searching for the best match in the local prior space (by optimizing the parameters encoding the local context) at the given query location. However, this requires the local context prior to generalize to a wide variety of unseen target regions, which is hard to achieve. To resolve this issue, we introduce Predictive Context Priors by learning Predictive Queries for each specific point cloud at inference time. Specifically, we first train a local context prior using a large point cloud dataset similar to previous techniques. For surface reconstruction at inference time, however, we specialize the local context prior into our Predictive Context Prior by learning Predictive Queries, which predict adjusted spatial query locations as displacements of the original locations. This leads to a global SDF that fits the specific point cloud the best. Intuitively, the query prediction enables us to flexibly search the learned local context prior over the entire prior space, rather than being restricted to the fixed query locations, and this improves the generalizability. Our method does not require ground truth signed distances, normals, or any additional procedure of signed distance fusion across overlapping regions. Our experimental results in surface reconstruction for single shapes or complex scenes show significant improvements over the state-of-the-art under widely used benchmarks.

65.Training and challenging models for text-guided fashion image retrieval ⬇️

Retrieving relevant images from a catalog based on a query image together with a modifying caption is a challenging multimodal task that can particularly benefit domains like apparel shopping, where fine details and subtle variations may be best expressed through natural language. We introduce a new evaluation dataset, Challenging Fashion Queries (CFQ), as well as a modeling approach that achieves state-of-the-art performance on the existing Fashion IQ (FIQ) dataset. CFQ complements existing benchmarks by including relative captions with positive and negative labels of caption accuracy and conditional image similarity, where others provided only positive labels with a combined meaning. We demonstrate the importance of multimodal pretraining for the task and show that domain-specific weak supervision based on attribute labels can augment generic large-scale pretraining. While previous modality fusion mechanisms lose the benefits of multimodal pretraining, we introduce a residual attention fusion mechanism that improves performance. We release CFQ and our code to the research community.

66.Cerebral Palsy Prediction with Frequency Attention Informed Graph Convolutional Networks ⬇️

Early diagnosis and intervention are clinically considered the paramount part of treating cerebral palsy (CP), so it is essential to design an efficient and interpretable automatic prediction system for CP. We highlight a significant difference between CP infants' frequency of human movement and that of the healthy group, which improves prediction performance. However, the existing deep learning-based methods did not use the frequency information of infants' movement for CP prediction. This paper proposes a frequency attention informed graph convolutional network and validates it on two consumer-grade RGB video datasets, namely MINI-RGBD and RVI-38 datasets. Our proposed frequency attention module aids in improving both classification performance and system interpretability. In addition, we design a frequency-binning method that retains the critical frequency of the human joint position data while filtering the noise. Our prediction performance achieves state-of-the-art research on both datasets. Our work demonstrates the effectiveness of frequency information in supporting the prediction of CP non-intrusively and provides a way for supporting the early diagnosis of CP in the resource-limited regions where the clinical resources are not abundant.

67.TerrainMesh: Metric-Semantic Terrain Reconstruction from Aerial Images Using Joint 2D-3D Learning ⬇️

This paper considers outdoor terrain mapping using RGB images obtained from an aerial vehicle. While feature-based localization and mapping techniques deliver real-time vehicle odometry and sparse keypoint depth reconstruction, a dense model of the environment geometry and semantics (vegetation, buildings, etc.) is usually recovered offline with significant computation and storage. This paper develops a joint 2D-3D learning approach to reconstruct a local metric-semantic mesh at each camera keyframe maintained by a visual odometry algorithm. Given the estimated camera trajectory, the local meshes can be assembled into a global environment model to capture the terrain topology and semantics during online operation. A local mesh is reconstructed using an initialization and refinement stage. In the initialization stage, we estimate the mesh vertex elevation by solving a least squares problem relating the vertex barycentric coordinates to the sparse keypoint depth measurements. In the refinement stage, we associate 2D image and semantic features with the 3D mesh vertices using camera projection and apply graph convolution to refine the mesh vertex spatial coordinates and semantic features based on joint 2D and 3D supervision. Quantitative and qualitative evaluation using real aerial images show the potential of our method to support environmental monitoring and surveillance applications.

68.Detecting Recolored Image by Spatial Correlation ⬇️

Image forensics, aiming to ensure the authenticity of the image, has made great progress in dealing with common image manipulation such as copy-move, splicing, and inpainting in the past decades. However, only a few researchers pay attention to an emerging editing technique called image recoloring, which can manipulate the color values of an image to give it a new style. To prevent it from being used maliciously, the previous approaches address the conventional recoloring from the perspective of inter-channel correlation and illumination consistency. In this paper, we try to explore a solution from the perspective of the spatial correlation, which exhibits the generic detection capability for both conventional and deep learning-based recoloring. Through theoretical and numerical analysis, we find that the recoloring operation will inevitably destroy the spatial correlation between pixels, implying a new prior of statistical discriminability. Based on such fact, we generate a set of spatial correlation features and learn the informative representation from the set via a convolutional neural network. To train our network, we use three recoloring methods to generate a large-scale and high-quality data set. Extensive experimental results in two recoloring scenes demonstrate that the spatial correlation features are highly discriminative. Our method achieves the state-of-the-art detection accuracy on multiple benchmark datasets and exhibits well generalization for unknown types of recoloring methods.

69.GRM: Gradient Rectification Module for Visual Place Retrieval ⬇️

Visual place retrieval aims to search images in the database that depict similar places as the query image. However, global descriptors encoded by the network usually fall into a low dimensional principal space, which is harmful to the retrieval performance. We first analyze the cause of this phenomenon, pointing out that it is due to degraded distribution of the gradients of descriptors. Then, a new module called Gradient Rectification Module(GRM) is proposed to alleviate this issue. It can be appended after the final pooling layer. This module can rectify the gradients to the complement space of the principal space. Therefore, the network is encouraged to generate descriptors more uniformly in the whole space. At last, we conduct experiments on multiple datasets and generalize our method to classification task under prototype learning framework.

70.Unsupervised Restoration of Weather-affected Images using Deep Gaussian Process-based CycleGAN ⬇️

Existing approaches for restoring weather-degraded images follow a fully-supervised paradigm and they require paired data for training. However, collecting paired data for weather degradations is extremely challenging, and existing methods end up training on synthetic data. To overcome this issue, we describe an approach for supervising deep networks that are based on CycleGAN, thereby enabling the use of unlabeled real-world data for training. Specifically, we introduce new losses for training CycleGAN that lead to more effective training, resulting in high-quality reconstructions. These new losses are obtained by jointly modeling the latent space embeddings of predicted clean images and original clean images through Deep Gaussian Processes. This enables the CycleGAN architecture to transfer the knowledge from one domain (weather-degraded) to another (clean) more effectively. We demonstrate that the proposed method can be effectively applied to different restoration tasks like de-raining, de-hazing and de-snowing and it outperforms other unsupervised techniques (that leverage weather-based characteristics) by a considerable margin.

71.CLIP-Dissect: Automatic Description of Neuron Representations in Deep Vision Networks ⬇️

In this paper, we propose CLIP-Dissect, a new technique to automatically describe the function of individual hidden neurons inside vision networks. CLIP-Dissect leverages recent advances in multimodal vision/language models to label internal neurons with open-ended concepts without the need for any labeled data or human examples, which are required for existing tools to succeed. We show that CLIP-Dissect provides more accurate descriptions than existing methods for neurons where the ground-truth is available as well as qualitatively good descriptions for hidden layer neurons. In addition, our method is very flexible: it is model agnostic, can easily handle new concepts and can be extended to take advantage of better multimodal models in the future. Finally CLIP-Dissect is computationally efficient and labels all neurons of a layer in a large vision model in tens of minutes.

72.Visual Attention Emerges from Recurrent Sparse Reconstruction ⬇️

Visual attention helps achieve robust perception under noise, corruption, and distribution shifts in human vision, which are areas where modern neural networks still fall short. We present VARS, Visual Attention from Recurrent Sparse reconstruction, a new attention formulation built on two prominent features of the human visual attention mechanism: recurrency and sparsity. Related features are grouped together via recurrent connections between neurons, with salient objects emerging via sparse regularization. VARS adopts an attractor network with recurrent connections that converges toward a stable pattern over time. Network layers are represented as ordinary differential equations (ODEs), formulating attention as a recurrent attractor network that equivalently optimizes the sparse reconstruction of input using a dictionary of "templates" encoding underlying patterns of data. We show that self-attention is a special case of VARS with a single-step optimization and no sparsity constraint. VARS can be readily used as a replacement for self-attention in popular vision transformers, consistently improving their robustness across various benchmarks. Code is released on GitHub (this https URL).

73.HRPlanes: High Resolution Airplane Dataset for Deep Learning ⬇️

Airplane detection from satellite imagery is a challenging task due to the complex backgrounds in the images and differences in data acquisition conditions caused by the sensor geometry and atmospheric effects. Deep learning methods provide reliable and accurate solutions for automatic detection of airplanes; however, huge amount of training data is required to obtain promising results. In this study, we create a novel airplane detection dataset called High Resolution Planes (HRPlanes) by using images from Google Earth (GE) and labeling the bounding box of each plane on the images. HRPlanes include GE images of several different airports across the world to represent a variety of landscape, seasonal and satellite geometry conditions obtained from different satellites. We evaluated our dataset with two widely used object detection methods namely YOLOv4 and Faster R-CNN. Our preliminary results show that the proposed dataset can be a valuable data source and benchmark data set for future applications. Moreover, proposed architectures and results of this study could be used for transfer learning of different datasets and models for airplane detection.

74.A Multi-level Alignment Training Scheme for Video-and-Language Grounding ⬇️

To solve video-and-language grounding tasks, the key is for the network to understand the connection between the two modalities. For a pair of video and language description, their semantic relation is reflected by their encodings' similarity. A good multi-modality encoder should be able to well capture both inputs' semantics and encode them in the shared feature space where embedding distance gets properly translated into their semantic similarity. In this work, we focused on this semantic connection between video and language, and developed a multi-level alignment training scheme to directly shape the encoding process. Global and segment levels of video-language alignment pairs were designed, based on the information similarity ranging from high-level context to fine-grained semantics. The contrastive loss was used to contrast the encodings' similarities between the positive and negative alignment pairs, and to ensure the network is trained in such a way that similar information is encoded closely in the shared feature space while information of different semantics is kept apart. Our multi-level alignment training can be applied to various video-and-language grounding tasks. Together with the task-specific training loss, our framework achieved comparable performance to previous state-of-the-arts on multiple video QA and retrieval datasets.

75.SegDiscover: Visual Concept Discovery via Unsupervised Semantic Segmentation ⬇️

Visual concept discovery has long been deemed important to improve interpretability of neural networks, because a bank of semantically meaningful concepts would provide us with a starting point for building machine learning models that exhibit intelligible reasoning process. Previous methods have disadvantages: either they rely on labelled support sets that incorporate human biases for objects that are "useful," or they fail to identify multiple concepts that occur within a single image. We reframe the concept discovery task as an unsupervised semantic segmentation problem, and present SegDiscover, a novel framework that discovers semantically meaningful visual concepts from imagery datasets with complex scenes without supervision. Our method contains three important pieces: generating concept primitives from raw images, discovering concepts by clustering in the latent space of a self-supervised pretrained encoder, and concept refinement via neural network smoothing. Experimental results provide evidence that our method can discover multiple concepts within a single image and outperforms state-of-the-art unsupervised methods on complex datasets such as Cityscapes and COCO-Stuff. Our method can be further used as a neural network explanation tool by comparing results obtained by different encoders.

76.Revealing Occlusions with 4D Neural Fields ⬇️

For computer vision systems to operate in dynamic situations, they need to be able to represent and reason about object permanence. We introduce a framework for learning to estimate 4D visual representations from monocular RGB-D, which is able to persist objects, even once they become obstructed by occlusions. Unlike traditional video representations, we encode point clouds into a continuous representation, which permits the model to attend across the spatiotemporal context to resolve occlusions. On two large video datasets that we release along with this paper, our experiments show that the representation is able to successfully reveal occlusions for several tasks, without any architectural changes. Visualizations show that the attention mechanism automatically learns to follow occluded objects. Since our approach can be trained end-to-end and is easily adaptable, we believe it will be useful for handling occlusions in many video understanding tasks. Data, code, and models are available at this https URL.

77.Label a Herd in Minutes: Individual Holstein-Friesian Cattle Identification ⬇️

We describe a practically evaluated approach for training visual cattle ID systems for a whole farm requiring only ten minutes of labelling effort. In particular, for the task of automatic identification of individual Holstein-Friesians in real-world farm CCTV, we show that self-supervision, metric learning, cluster analysis, and active learning can complement each other to significantly reduce the annotation requirements usually needed to train cattle identification frameworks. Evaluating the approach on the test portion of the publicly available Cows2021 dataset, for training we use 23,350 frames across 435 single individual tracklets generated by automated oriented cattle detection and tracking in operational farm footage. Self-supervised metric learning is first employed to initialise a candidate identity space where each tracklet is considered a distinct entity. Grouping entities into equivalence classes representing cattle identities is then performed by automated merging via cluster analysis and active learning. Critically, we identify the inflection point at which automated choices cannot replicate improvements based on human intervention to reduce annotation to a minimum. Experimental results show that cluster analysis and a few minutes of labelling after automated self-supervision can improve the test identification accuracy of 153 identities to 92.44% (ARI=0.93) from the 74.9% (ARI=0.754) obtained by self-supervision only. These promising results indicate that a tailored combination of human and machine reasoning in visual cattle ID pipelines can be highly effective whilst requiring only minimal labelling effort. We provide all key source code and network weights with this paper for easy result reproduction.

78.Transfer Learning from Synthetic In-vitro Soybean Pods Dataset for In-situ Segmentation of On-branch Soybean Pod ⬇️

The mature soybean plants are of complex architecture with pods frequently touching each other, posing a challenge for in-situ segmentation of on-branch soybean pods. Deep learning-based methods can achieve accurate training and strong generalization capabilities, but it demands massive labeled data, which is often a limitation, especially for agricultural applications. As lacking the labeled data to train an in-situ segmentation model for on-branch soybean pods, we propose a transfer learning from synthetic in-vitro soybean pods. First, we present a novel automated image generation method to rapidly generate a synthetic in-vitro soybean pods dataset with plenty of annotated samples. The in-vitro soybean pods samples are overlapped to simulate the frequently physically touching of on-branch soybean pods. Then, we design a two-step transfer learning. In the first step, we finetune an instance segmentation network pretrained by a source domain (MS COCO dataset) with a synthetic target domain (in-vitro soybean pods dataset). In the second step, transferring from simulation to reality is performed by finetuning on a few real-world mature soybean plant samples. The experimental results show the effectiveness of the proposed two-step transfer learning method, such that AP${50}$ was 0.80 for the real-world mature soybean plant test dataset, which is higher than that of direct adaptation and its AP${50}$ was 0.77. Furthermore, the visualizations of in-situ segmentation results of on-branch soybean pods show that our method performs better than other methods, especially when soybean pods overlap densely.

79.Generative sampling in tractography using autoencoders (GESTA) ⬇️

Current tractography methods use the local orientation information to propagate streamlines from seed locations. Many such seeds provide streamlines that stop prematurely or fail to map the true pathways because some white matter bundles are "harder-to-track" than others. This results in tractography reconstructions with poor white and gray matter spatial coverage. In this work, we propose a generative, autoencoder-based method, named GESTA (Generative Sampling in Tractography using Autoencoders), that produces streamlines with better spatial coverage. Compared to other deep learning methods, our autoencoder-based framework is not constrained by any prior or a fixed set of bundles. GESTA produces new and complete streamlines for any white matter bundle. GESTA is shown to be effective on both synthetic and human brain in vivo data. Our streamline evaluation framework ensures that the streamlines produced by GESTA are anatomically plausible and fit well to the local diffusion signal. The streamline evaluation criteria assess anatomy (white matter coverage), local orientation alignment (direction), geometry features of streamlines, and optionally, gray matter connectivity. The GESTA framework offers considerable gains in bundle coverage using a reduced set of seeding streamlines with a 1.5x improvement for the "Fiber Cup", and 6x for the ISMRM 2015 Tractography Challenge datasets. Similarly, it provides a 4x white matter volume increase on the BIL&GIN callosal homotopic dataset. It also successfully generates new streamlines in poorly populated bundles, such as the fornix and other hard-to-track bundles, on in vivo data. GESTA is thus the first deep tractography generative method that can improve white matter reconstruction of hard-to-track bundles.

80.Identity Preserving Loss for Learned Image Compression ⬇️

Deep learning model inference on embedded devices is challenging due to the limited availability of computation resources. A popular alternative is to perform model inference on the cloud, which requires transmitting images from the embedded device to the cloud. Image compression techniques are commonly employed in such cloud-based architectures to reduce transmission latency over low bandwidth networks. This work proposes an end-to-end image compression framework that learns domain-specific features to achieve higher compression ratios than standard HEVC/JPEG compression techniques while maintaining accuracy on downstream tasks (e.g., recognition). Our framework does not require fine-tuning of the downstream task, which allows us to drop-in any off-the-shelf downstream task model without retraining. We choose faces as an application domain due to the ready availability of datasets and off-the-shelf recognition models as representative downstream tasks. We present a novel Identity Preserving Reconstruction (IPR) loss function which achieves Bits-Per-Pixel (BPP) values that are ~38% and ~42% of CRF-23 HEVC compression for LFW (low-resolution) and CelebA-HQ (high-resolution) datasets, respectively, while maintaining parity in recognition accuracy. The superior compression ratio is achieved as the model learns to retain the domain-specific features (e.g., facial features) while sacrificing details in the background. Furthermore, images reconstructed by our proposed compression model are robust to changes in downstream model architectures. We show at-par recognition performance on the LFW dataset with an unseen recognition model while retaining a lower BPP value of ~38% of CRF-23 HEVC compression.

81.Performer: A Novel PPG to ECG Reconstruction Transformer For a Digital Biomarker of Cardiovascular Disease Detection ⬇️

Cardiovascular diseases (CVDs) have become the top one cause of death; three-quarters of these deaths occur in lower-income communities. Electrocardiography (ECG), an electrical measurement capturing the cardiac activities, is a gold-standard to diagnose CVDs. However, ECG is infeasible for continuous cardiac monitoring due to its requirement for user participation. Meanwhile, photoplethysmography (PPG) is easy to collect, but the limited accuracy constrains its clinical usage. In this research, a novel Transformer-based architecture, Performer, is invented to reconstruct ECG from PPG and to create a novel digital biomarker, PPG along with its reconstructed ECG, as multiple modalities for CVD detection. This architecture, for the first time, performs Transformer sequence to sequence translation on biomedical waveforms, while also utilizing the advantages of the easily accessible PPG and the well-studied base of ECG. Shifted Patch-based Attention (SPA) is created to maximize the signal features by fetching the various sequence lengths as hierarchical stages into the training while also capturing cross-patch connections through the shifted patch mechanism. This architecture generates a state-of-the-art performance of 0.29 RMSE for reconstructing ECG from PPG, achieving an average of 95.9% diagnosis for CVDs on the MIMIC III dataset and 75.9% for diabetes on the PPG-BP dataset. Performer, along with its novel digital biomarker, offers a low-cost and non-invasive solution for continuous cardiac monitoring, only requiring the easily extractable PPG data to reconstruct the not-as-accessible ECG data. As a prove of concept, an earring wearable, named PEARL (prototype), is designed to scale up the point-of-care (POC) healthcare system.

82.Enable Deep Learning on Mobile Devices: Methods, Systems, and Applications ⬇️

Deep neural networks (DNNs) have achieved unprecedented success in the field of artificial intelligence (AI), including computer vision, natural language processing and speech recognition. However, their superior performance comes at the considerable cost of computational complexity, which greatly hinders their applications in many resource-constrained devices, such as mobile phones and Internet of Things (IoT) devices. Therefore, methods and techniques that are able to lift the efficiency bottleneck while preserving the high accuracy of DNNs are in great demand in order to enable numerous edge AI applications. This paper provides an overview of efficient deep learning methods, systems and applications. We start from introducing popular model compression methods, including pruning, factorization, quantization as well as compact model design. To reduce the large design cost of these manual solutions, we discuss the AutoML framework for each of them, such as neural architecture search (NAS) and automated pruning and quantization. We then cover efficient on-device training to enable user customization based on the local data on mobile devices. Apart from general acceleration techniques, we also showcase several task-specific accelerations for point cloud, video and natural language processing by exploiting their spatial sparsity and temporal/token redundancy. Finally, to support all these algorithmic advancements, we introduce the efficient deep learning system design from both software and hardware perspectives.

83.CellDefectNet: A Machine-designed Attention Condenser Network for Electroluminescence-based Photovoltaic Cell Defect Inspection ⬇️

Photovoltaic cells are electronic devices that convert light energy to electricity, forming the backbone of solar energy harvesting systems. An essential step in the manufacturing process for photovoltaic cells is visual quality inspection using electroluminescence imaging to identify defects such as cracks, finger interruptions, and broken cells. A big challenge faced by industry in photovoltaic cell visual inspection is the fact that it is currently done manually by human inspectors, which is extremely time consuming, laborious, and prone to human error. While deep learning approaches holds great potential to automating this inspection, the hardware resource-constrained manufacturing scenario makes it challenging for deploying complex deep neural network architectures. In this work, we introduce CellDefectNet, a highly efficient attention condenser network designed via machine-driven design exploration specifically for electroluminesence-based photovoltaic cell defect detection on the edge. We demonstrate the efficacy of CellDefectNet on a benchmark dataset comprising of a diversity of photovoltaic cells captured using electroluminescence imagery, achieving an accuracy of ~86.3% while possessing just 410K parameters (~13$\times$ lower than EfficientNet-B0, respectively) and ~115M FLOPs (~12$\times$ lower than EfficientNet-B0) and ~13$\times$ faster on an ARM Cortex A-72 embedded processor when compared to EfficientNet-B0.

84.LightDefectNet: A Highly Compact Deep Anti-Aliased Attention Condenser Neural Network Architecture for Light Guide Plate Surface Defect Detection ⬇️

Light guide plates are essential optical components widely used in a diverse range of applications ranging from medical lighting fixtures to back-lit TV displays. An essential step in the manufacturing of light guide plates is the quality inspection of defects such as scratches, bright/dark spots, and impurities. This is mainly done in industry through manual visual inspection for plate pattern irregularities, which is time-consuming and prone to human error and thus act as a significant barrier to high-throughput production. Advances in deep learning-driven computer vision has led to the exploration of automated visual quality inspection of light guide plates to improve inspection consistency, accuracy, and efficiency. However, given the cost constraints in visual inspection scenarios, the widespread adoption of deep learning-driven computer vision methods for inspecting light guide plates has been greatly limited due to high computational requirements. In this study, we explore the utilization of machine-driven design exploration with computational and "best-practices" constraints as well as L$_1$ paired classification discrepancy loss to create LightDefectNet, a highly compact deep anti-aliased attention condenser neural network architecture tailored specifically for light guide plate surface defect detection in resource-constrained scenarios. Experiments show that LightDetectNet achieves a detection accuracy of $\sim$98.2% on the LGPSDD benchmark while having just 770K parameters ($\sim$33$\times$ and $\sim$6.9$\times$ lower than ResNet-50 and EfficientNet-B0, respectively) and $\sim$93M FLOPs ($\sim$88$\times$ and $\sim$8.4$\times$ lower than ResNet-50 and EfficientNet-B0, respectively) and $\sim$8.8$\times$ faster inference speed than EfficientNet-B0 on an embedded ARM processor.

85.Do Users Benefit From Interpretable Vision? A User Study, Baseline, And Dataset ⬇️

A variety of methods exist to explain image classification models. However, whether they provide any benefit to users over simply comparing various inputs and the model's respective predictions remains unclear. We conducted a user study (N=240) to test how such a baseline explanation technique performs against concept-based and counterfactual explanations. To this end, we contribute a synthetic dataset generator capable of biasing individual attributes and quantifying their relevance to the model. In a study, we assess if participants can identify the relevant set of attributes compared to the ground-truth. Our results show that the baseline outperformed concept-based explanations. Counterfactual explanations from an invertible neural network performed similarly as the baseline. Still, they allowed users to identify some attributes more accurately. Our results highlight the importance of measuring how well users can reason about biases of a model, rather than solely relying on technical evaluations or proxy tasks. We open-source our study and dataset so it can serve as a blue-print for future studies. For code see, this https URL

86.Translating Clinical Delineation of Diabetic Foot Ulcers into Machine Interpretable Segmentation ⬇️

Diabetic foot ulcer is a severe condition that requires close monitoring and management. For training machine learning methods to auto-delineate the ulcer, clinical staff must provide ground truth annotations. In this paper, we propose a new diabetic foot ulcers dataset, namely DFUC2022, the largest segmentation dataset where ulcer regions were manually delineated by clinicians. We assess whether the clinical delineations are machine interpretable by deep learning networks or if image processing refined contour should be used. By providing benchmark results using a selection of popular deep learning algorithms, we draw new insights into the limitations of DFU wound delineation and report on the associated issues. This paper provides some observations on baseline models to facilitate DFUC2022 Challenge in conjunction with MICCAI 2022. The leaderboard will be ranked by Dice score, where the best FCN-based method is 0.5708 and DeepLabv3+ achieved the best score of 0.6277. This paper demonstrates that image processing using refined contour as ground truth can provide better agreement with machine predicted results. DFUC2022 will be released on the 27th April 2022.

87.Deflectometry for specular surfaces: an overview ⬇️

Deflectometry as a technical approach to assessing reflective surfaces has now existed for almost 40 years. Different aspects and variations of the method have been studied in multiple theses and research articles, and reviews are also becoming available for certain subtopics. Still a field of active development with many unsolved problems, deflectometry now encompasses a large variety of application domains, hardware setup types, and processing workflows designed for different purposes, and spans a range from qualitative defect inspection of large vehicles to precision measurements of microscopic optics. Over these years, many exciting developments have accumulated in the underlying theory, in the systems design, and in the implementation specifics. This diversity of topics is difficult to grasp for experts and non-experts alike and may present an obstacle to a wider acceptance of deflectometry as a useful tool in other research fields and in the industry.
This paper presents an attempt to summarize the status of deflectometry, and to map relations between its notable "spin-off" branches. The intention of the paper is to provide a common communication basis for practitioners and at the same time to offer a convenient entry point for those interested in learning and using the method. The list of references is extensive but definitely not exhaustive, introducing some prominent trends and established research groups in order to facilitate further self-directed exploration by the reader.

88.Ad Creative Discontinuation Prediction with Multi-Modal Multi-Task Neural Survival Networks ⬇️

Discontinuing ad creatives at an appropriate time is one of the most important ad operations that can have a significant impact on sales. Such operational support for ineffective ads has been less explored than that for effective ads. After pre-analyzing 1,000,000 real-world ad creatives, we found that there are two types of discontinuation: short-term (i.e., cut-out) and long-term (i.e., wear-out). In this paper, we propose a practical prediction framework for the discontinuation of ad creatives with a hazard function-based loss function inspired by survival analysis. Our framework predicts the discontinuations with a multi-modal deep neural network that takes as input the ad creative (e.g., text, categorical, image, numerical features). To improve the prediction performance for the two different types of discontinuations and for the ad creatives that contribute to sales, we introduce two new techniques: (1) a two-term estimation technique with multi-task learning and (2) a click-through rate-weighting technique for the loss function. We evaluated our framework using the large-scale ad creative dataset, including 10 billion scale impressions. In terms of the concordance index (short: 0.896, long: 0.939, and overall: 0.792), our framework achieved significantly better performance than the conventional method (0.531). Additionally, we confirmed that our framework (i) demonstrated the same degree of discontinuation effect as manual operations for short-term cases, and (ii) accurately predicted the ad discontinuation order, which is important for long-running ad creatives for long-term cases.

89.Faculty Distillation with Optimal Transport ⬇️

Knowledge distillation (KD) has shown its effectiveness in improving a student classifier given a suitable teacher. The outpouring of diverse and plentiful pre-trained models may provide abundant teacher resources for KD. However, these models are often trained on different tasks from the student, which requires the student to precisely select the most contributive teacher and enable KD across different label spaces. These restrictions disclose the insufficiency of standard KD and motivate us to study a new paradigm called faculty distillation. Given a group of teachers (faculty), a student needs to select the most relevant teacher and perform generalized knowledge reuse. To this end, we propose to link teacher's task and student's task by optimal transport. Based on the semantic relationship between their label spaces, we can bridge the support gap between output distributions by minimizing Sinkhorn distances. The transportation cost also acts as a measurement of teachers' adaptability so that we can rank the teachers efficiently according to their relatedness. Experiments under various settings demonstrate the succinctness and versatility of our method.

90.Multimodal Dual Emotion with Fusion of Visual Sentiment for Rumor Detection ⬇️

In recent years, rumors have had a devastating impact on society, making rumor detection a significant challenge. However, the studies on rumor detection ignore the intense emotions of images in the rumor content. This paper verifies that the image emotion improves the rumor detection efficiency. A Multimodal Dual Emotion feature in rumor detection, which consists of visual and textual emotions, is proposed. To the best of our knowledge, this is the first study which uses visual emotion in rumor detection. The experiments on real datasets verify that the proposed features outperform the state-of-the-art sentiment features, and can be extended in rumor detectors while improving their performance.

91.Self-supervision versus synthetic datasets: which is the lesser evil in the context of video denoising? ⬇️

Supervised training has led to state-of-the-art results in image and video denoising. However, its application to real data is limited since it requires large datasets of noisy-clean pairs that are difficult to obtain. For this reason, networks are often trained on realistic synthetic data. More recently, some self-supervised frameworks have been proposed for training such denoising networks directly on the noisy data without requiring ground truth. On synthetic denoising problems supervised training outperforms self-supervised approaches, however in recent years the gap has become narrower, especially for video. In this paper, we propose a study aiming to determine which is the best approach to train denoising networks for real raw videos: supervision on synthetic realistic data or self-supervision on real data. A complete study with quantitative results in case of natural videos with real motion is impossible since no dataset with clean-noisy pairs exists. We address this issue by considering three independent experiments in which we compare the two frameworks. We found that self-supervision on the real data outperforms supervision on synthetic data, and that in normal illumination conditions the drop in performance is due to the synthetic ground truth generation, not the noise model.

92.End-to-End Audio Strikes Back: Boosting Augmentations Towards An Efficient Audio Classification Network ⬇️

While efficient architectures and a plethora of augmentations for end-to-end image classification tasks have been suggested and heavily investigated, state-of-the-art techniques for audio classifications still rely on numerous representations of the audio signal together with large architectures, finetuned from large datasets. By utilizing the inherited lightweight nature of audio and novel audio augmentations, we were able to present an efficient end-to-end1 network with strong generalization ability. Experiments on a variety of sound classification sets demonstrate the effectiveness and robustness of our approach, by achieving state-of-the-art results in various settings. Public code will be available.

93.High-Efficiency Lossy Image Coding Through Adaptive Neighborhood Information Aggregation ⬇️

Questing for lossy image coding (LIC) with superior efficiency on both compression performance and computation throughput is challenging. The vital factor behind is how to intelligently explore Adaptive Neighborhood Information Aggregation (ANIA) in transform and entropy coding modules. To this aim, Integrated Convolution and Self-Attention (ICSA) unit is first proposed to form content-adaptive transform to dynamically characterize and embed neighborhood information conditioned on the input. Then a Multistage Context Model (MCM) is developed to stagewisely execute context prediction using necessary neighborhood elements for accurate and parallel entropy probability estimation. Both ICSA and MCM are stacked under a Variational Auto-Encoder (VAE) architecture to derive rate-distortion optimized compact representation of input image via end-to-end training. Our method reports the superior compression performance surpassing the VVC Intra with $\approx$15% BD-rate improvement averaged across Kodak, CLIC and Tecnick datasets; and also demonstrates $\approx$10$\times$ speedup of image decoding when compared with other notable learned LIC approaches. All materials are made publicly accessible at this https URL for reproducible research.

94.BCI: Breast Cancer Immunohistochemical Image Generation through Pyramid Pix2pix ⬇️

The evaluation of human epidermal growth factor receptor 2 (HER2) expression is essential to formulate a precise treatment for breast cancer. The routine evaluation of HER2 is conducted with immunohistochemical techniques (IHC), which is very expensive. Therefore, for the first time, we propose a breast cancer immunohistochemical (BCI) benchmark attempting to synthesize IHC data directly with the paired hematoxylin and eosin (HE) stained images. The dataset contains 4870 registered image pairs, covering a variety of HER2 expression levels. Based on BCI, as a minor contribution, we further build a pyramid pix2pix image generation method, which achieves better HE to IHC translation results than the other current popular algorithms. Extensive experiments demonstrate that BCI poses new challenges to the existing image translation research. Besides, BCI also opens the door for future pathology studies in HER2 expression evaluation based on the synthesized IHC images. BCI dataset can be downloaded from this https URL.

95.Tensorial tomographic differential phase-contrast microscopy ⬇️

We report Tensorial Tomographic Differential Phase-Contrast microscopy (T2DPC), a quantitative label-free tomographic imaging method for simultaneous measurement of phase and anisotropy. T2DPC extends differential phase-contrast microscopy, a quantitative phase imaging technique, to highlight the vectorial nature of light. The method solves for permittivity tensor of anisotropic samples from intensity measurements acquired with a standard microscope equipped with an LED matrix, a circular polarizer, and a polarization-sensitive camera. We demonstrate accurate volumetric reconstructions of refractive index, birefringence, and orientation for various validation samples, and show that the reconstructed polarization structures of a biological specimen are predictive of pathology.

96.Deep Learning for Medical Image Registration: A Comprehensive Review ⬇️

Image registration is a critical component in the applications of various medical image analyses. In recent years, there has been a tremendous surge in the development of deep learning (DL)-based medical image registration models. This paper provides a comprehensive review of medical image registration. Firstly, a discussion is provided for supervised registration categories, for example, fully supervised, dual supervised, and weakly supervised registration. Next, similarity-based as well as generative adversarial network (GAN)-based registration are presented as part of unsupervised registration. Deep iterative registration is then described with emphasis on deep similarity-based and reinforcement learning-based registration. Moreover, the application areas of medical image registration are reviewed. This review focuses on monomodal and multimodal registration and associated imaging, for instance, X-ray, CT scan, ultrasound, and MRI. The existing challenges are highlighted in this review, where it is shown that a major challenge is the absence of a training dataset with known transformations. Finally, a discussion is provided on the promising future research areas in the field of DL-based medical image registration.

97.Dictionary Attacks on Speaker Verification ⬇️

In this paper, we propose dictionary attacks against speaker verification - a novel attack vector that aims to match a large fraction of speaker population by chance. We introduce a generic formulation of the attack that can be used with various speech representations and threat models. The attacker uses adversarial optimization to maximize raw similarity of speaker embeddings between a seed speech sample and a proxy population. The resulting master voice successfully matches a non-trivial fraction of people in an unknown population. Adversarial waveforms obtained with our approach can match on average 69% of females and 38% of males enrolled in the target system at a strict decision threshold calibrated to yield false alarm rate of 1%. By using the attack with a black-box voice cloning system, we obtain master voices that are effective in the most challenging conditions and transferable between speaker encoders. We also show that, combined with multiple attempts, this attack opens even more to serious issues on the security of these systems.

98.Colorectal cancer survival prediction using deep distribution based multiple-instance learning ⬇️

Several deep learning algorithms have been developed to predict survival of cancer patients using whole slide images (WSIs).However, identification of image phenotypes within the WSIs that are relevant to patient survival and disease progression is difficult for both clinicians, and deep learning algorithms. Most deep learning based Multiple Instance Learning (MIL) algorithms for survival prediction use either top instances (e.g., maxpooling) or top/bottom instances (e.g., MesoNet) to identify image phenotypes. In this study, we hypothesize that wholistic information of the distribution of the patch scores within a WSI can predict the cancer survival better. We developed a distribution based multiple-instance survival learning algorithm (DeepDisMISL) to validate this hypothesis. We designed and executed experiments using two large international colorectal cancer WSIs datasets - MCO CRC and TCGA COAD-READ. Our results suggest that the more information about the distribution of the patch scores for a WSI, the better is the prediction performance. Including multiple neighborhood instances around each selected distribution location (e.g., percentiles) could further improve the prediction. DeepDisMISL demonstrated superior predictive ability compared to other recently published, state-of-the-art algorithms. Furthermore, our algorithm is interpretable and could assist in understanding the relationship between cancer morphological phenotypes and patients cancer survival risk.

99.PUERT: Probabilistic Under-sampling and Explicable Reconstruction Network for CS-MRI ⬇️

Compressed Sensing MRI (CS-MRI) aims at reconstructing de-aliased images from sub-Nyquist sampling k-space data to accelerate MR Imaging, thus presenting two basic issues, i.e., where to sample and how to reconstruct. To deal with both problems simultaneously, we propose a novel end-to-end Probabilistic Under-sampling and Explicable Reconstruction neTwork, dubbed PUERT, to jointly optimize the sampling pattern and the reconstruction network. Instead of learning a deterministic mask, the proposed sampling subnet explores an optimal probabilistic sub-sampling pattern, which describes independent Bernoulli random variables at each possible sampling point, thus retaining robustness and stochastics for a more reliable CS reconstruction. A dynamic gradient estimation strategy is further introduced to gradually approximate the binarization function in backward propagation, which efficiently preserves the gradient information and further improves the reconstruction quality. Moreover, in our reconstruction subnet, we adopt a model-based network design scheme with high efficiency and interpretability, which is shown to assist in further exploitation for the sampling subnet. Extensive experiments on two widely used MRI datasets demonstrate that our proposed PUERT not only achieves state-of-the-art results in terms of both quantitative metrics and visual quality but also yields a sub-sampling pattern and a reconstruction model that are both customized to training data.

100.Realistic Evaluation of Transductive Few-Shot Learning ⬇️

Transductive inference is widely used in few-shot learning, as it leverages the statistics of the unlabeled query set of a few-shot task, typically yielding substantially better performances than its inductive counterpart. The current few-shot benchmarks use perfectly class-balanced tasks at inference. We argue that such an artificial regularity is unrealistic, as it assumes that the marginal label probability of the testing samples is known and fixed to the uniform distribution. In fact, in realistic scenarios, the unlabeled query sets come with arbitrary and unknown label marginals. We introduce and study the effect of arbitrary class distributions within the query sets of few-shot tasks at inference, removing the class-balance artefact. Specifically, we model the marginal probabilities of the classes as Dirichlet-distributed random variables, which yields a principled and realistic sampling within the simplex. This leverages the current few-shot benchmarks, building testing tasks with arbitrary class distributions. We evaluate experimentally state-of-the-art transductive methods over 3 widely used data sets, and observe, surprisingly, substantial performance drops, even below inductive methods in some cases. Furthermore, we propose a generalization of the mutual-information loss, based on $\alpha$-divergences, which can handle effectively class-distribution variations. Empirically, we show that our transductive $\alpha$-divergence optimization outperforms state-of-the-art methods across several data sets, models and few-shot settings. Our code is publicly available at this https URL.

101.Gabor is Enough: Interpretable Deep Denoising with a Gabor Synthesis Dictionary Prior ⬇️

Image processing neural networks, natural and artificial, have a long history with orientation-selectivity, often described mathematically as Gabor filters. Gabor-like filters have been observed in the early layers of CNN classifiers and even throughout low-level image processing networks. In this work, we take this observation to the extreme and explicitly constrain the filters of a natural-image denoising CNN to be learned 2D real Gabor filters. Surprisingly, we find that the proposed network (GDLNet) can achieve near state-of-the-art denoising performance amongst popular fully convolutional neural networks, with only a fraction of the learned parameters. We further verify that this parameterization maintains the noise-level generalization (training vs. inference mismatch) characteristics of the base network, and investigate the contribution of individual Gabor filter parameters to the performance of the denoiser. We present positive findings for the interpretation of dictionary learning networks as performing accelerated sparse-coding via the importance of untied learned scale parameters between network layers. Our network's success suggests that representations used by low-level image processing CNNs can be as simple and interpretable as Gabor filterbanks.

102.Towards Bundle Adjustment for Satellite Imaging via Quantum Machine Learning ⬇️

Given is a set of images, where all images show views of the same area at different points in time and from different viewpoints. The task is the alignment of all images such that relevant information, e.g., poses, changes, and terrain, can be extracted from the fused image. In this work, we focus on quantum methods for keypoint extraction and feature matching, due to the demanding computational complexity of these sub-tasks. To this end, k-medoids clustering, kernel density clustering, nearest neighbor search, and kernel methods are investigated and it is explained how these methods can be re-formulated for quantum annealers and gate-based quantum computers. Experimental results obtained on digital quantum emulation hardware, quantum annealers, and quantum gate computers show that classical systems still deliver superior results. However, the proposed methods are ready for the current and upcoming generations of quantum computing devices which have the potential to outperform classical systems in the near future.

103.Transformation Invariant Cancerous Tissue Classification Using Spatially Transformed DenseNet ⬇️

In this work, we introduce a spatially transformed DenseNet architecture for transformation invariant classification of cancer tissue. Our architecture increases the accuracy of the base DenseNet architecture while adding the ability to operate in a transformation invariant way while simultaneously being simpler than other models that try to provide some form of invariance.

104.Class Balanced PixelNet for Neurological Image Segmentation ⬇️

In this paper, we propose an automatic brain tumor segmentation approach (e.g., PixelNet) using a pixel-level convolutional neural network (CNN). The model extracts feature from multiple convolutional layers and concatenate them to form a hyper-column where samples a modest number of pixels for optimization. Hyper-column ensures both local and global contextual information for pixel-wise predictors. The model confirms the statistical efficiency by sampling a few pixels in the training phase where spatial redundancy limits the information learning among the neighboring pixels in conventional pixel-level semantic segmentation approaches. Besides, label skewness in training data leads the convolutional model often converge to certain classes which is a common problem in the medical dataset. We deal with this problem by selecting an equal number of pixels for all the classes in sampling time. The proposed model has achieved promising results in brain tumor and ischemic stroke lesion segmentation datasets.

105.Towards Data-Free Model Stealing in a Hard Label Setting ⬇️

Machine learning models deployed as a service (MLaaS) are susceptible to model stealing attacks, where an adversary attempts to steal the model within a restricted access framework. While existing attacks demonstrate near-perfect clone-model performance using softmax predictions of the classification network, most of the APIs allow access to only the top-1 labels. In this work, we show that it is indeed possible to steal Machine Learning models by accessing only top-1 predictions (Hard Label setting) as well, without access to model gradients (Black-Box setting) or even the training dataset (Data-Free setting) within a low query budget. We propose a novel GAN-based framework that trains the student and generator in tandem to steal the model effectively while overcoming the challenge of the hard label setting by utilizing gradients of the clone network as a proxy to the victim's gradients. We propose to overcome the large query costs associated with a typical Data-Free setting by utilizing publicly available (potentially unrelated) datasets as a weak image prior. We additionally show that even in the absence of such data, it is possible to achieve state-of-the-art results within a low query budget using synthetically crafted samples. We are the first to demonstrate the scalability of Model Stealing in a restricted access setting on a 100 class dataset as well.

106.Indoor simultaneous localization and mapping based on fringe projection profilometry ⬇️

Simultaneous Localization and Mapping (SLAM) plays an important role in outdoor and indoor applications ranging from autonomous driving to indoor robotics. Outdoor SLAM has been widely used with the assistance of LiDAR or GPS. For indoor applications, the LiDAR technique does not satisfy the accuracy requirement and the GPS signals will be lost. An accurate and efficient scene sensing technique is required for indoor SLAM. As the most promising 3D sensing technique, the opportunities for indoor SLAM with fringe projection profilometry (FPP) systems are obvious, but methods to date have not fully leveraged the accuracy and speed of sensing that such systems offer. In this paper, we propose a novel FPP-based indoor SLAM method based on the coordinate transformation relationship of FPP, where the 2D-to-3D descriptor-assisted is used for mapping and localization. The correspondences generated by matching descriptors are used for fast and accurate mapping, and the transform estimation between the 2D and 3D descriptors is used to localize the sensor. The provided experimental results demonstrate that the proposed indoor SLAM can achieve the localization and mapping accuracy around one millimeter.

107.Federated Contrastive Learning for Volumetric Medical Image Segmentation ⬇️

Supervised deep learning needs a large amount of labeled data to achieve high performance. However, in medical imaging analysis, each site may only have a limited amount of data and labels, which makes learning ineffective. Federated learning (FL) can help in this regard by learning a shared model while keeping training data local for privacy. Traditional FL requires fully-labeled data for training, which is inconvenient or sometimes infeasible to obtain due to high labeling cost and the requirement of expertise. Contrastive learning (CL), as a self-supervised learning approach, can effectively learn from unlabeled data to pre-train a neural network encoder, followed by fine-tuning for downstream tasks with limited annotations. However, when adopting CL in FL, the limited data diversity on each client makes federated contrastive learning (FCL) ineffective. In this work, we propose an FCL framework for volumetric medical image segmentation with limited annotations. More specifically, we exchange the features in the FCL pre-training process such that diverse contrastive data are provided to each site for effective local CL while keeping raw data private. Based on the exchanged features, global structural matching further leverages the structural similarity to align local features to the remote ones such that a unified feature space can be learned among different sites. Experiments on a cardiac MRI dataset show the proposed framework substantially improves the segmentation performance compared with state-of-the-art techniques.

108.Evaluation of Multi-Scale Multiple Instance Learning to Improve Thyroid Cancer Classification ⬇️

Thyroid cancer is currently the fifth most common malignancy diagnosed in women. Since differentiation of cancer sub-types is important for treatment and current, manual methods are time consuming and subjective, automatic computer-aided differentiation of cancer types is crucial. Manual differentiation of thyroid cancer is based on tissue sections, analysed by pathologists using histological features. Due to the enormous size of gigapixel whole slide images, holistic classification using deep learning methods is not feasible. Patch based multiple instance learning approaches, combined with aggregations such as bag-of-words, is a common approach. This work's contribution is to extend a patch based state-of-the-art method by generating and combining feature vectors of three different patch resolutions and analysing three distinct ways of combining them. The results showed improvements in one of the three multi-scale approaches, while the others led to decreased scores. This provides motivation for analysis and discussion of the individual approaches.

109.Unified Pretraining Framework for Document Understanding ⬇️

Document intelligence automates the extraction of information from documents and supports many business applications. Recent self-supervised learning methods on large-scale unlabeled document datasets have opened up promising directions towards reducing annotation efforts by training models with self-supervised objectives. However, most of the existing document pretraining methods are still language-dominated. We present UDoc, a new unified pretraining framework for document understanding. UDoc is designed to support most document understanding tasks, extending the Transformer to take multimodal embeddings as input. Each input element is composed of words and visual features from a semantic region of the input document image. An important feature of UDoc is that it learns a generic representation by making use of three self-supervised losses, encouraging the representation to model sentences, learn similarities, and align modalities. Extensive empirical analysis demonstrates that the pretraining procedure learns better joint representations and leads to improvements in downstream tasks.