1.Deep Kernelized Dense Geometric Matching ⬇️
Dense geometric matching is a challenging computer vision task, requiring accurate correspondences under extreme variations in viewpoint and illumination, even for low-texture regions. In this task, finding accurate global correspondences is essential for later refinement stages. The current learning based paradigm is to perform global fixed-size correlation, followed by flattening and convolution to predict correspondences. In this work, we consider the problem from a different perspective and propose to formulate global correspondence estimation as a continuous probabilistic regression task using deep kernels, yielding a novel approach to learning dense correspondences. Our full approach, \textbf{D}eep \textbf{K}ernelized \textbf{M}atching, achieves significant improvements compared to the state-of-the-art on the competitive HPatches and YFCC100m benchmarks, and we dissect the gains of our contributions in a thorough ablation study.
2.Interactron: Embodied Adaptive Object Detection ⬇️
Over the years various methods have been proposed for the problem of object detection. Recently, we have witnessed great strides in this domain owing to the emergence of powerful deep neural networks. However, there are typically two main assumptions common among these approaches. First, the model is trained on a fixed training set and is evaluated on a pre-recorded test set. Second, the model is kept frozen after the training phase, so no further updates are performed after the training is finished. These two assumptions limit the applicability of these methods to real-world settings. In this paper, we propose Interactron, a method for adaptive object detection in an interactive setting, where the goal is to perform object detection in images observed by an embodied agent navigating in different environments. Our idea is to continue training during inference and adapt the model at test time without any explicit supervision via interacting with the environment. Our adaptive object detection model provides a 11.8 point improvement in AP (and 19.1 points in AP50) over DETR, a recent, high-performance object detector. Moreover, we show that our object detection model adapts to environments with completely different appearance characteristics, and its performance is on par with a model trained with full supervision for those environments.
3.Stay Positive: Non-Negative Image Synthesis for Augmented Reality ⬇️
In applications such as optical see-through and projector augmented reality, producing images amounts to solving non-negative image generation, where one can only add light to an existing image. Most image generation methods, however, are ill-suited to this problem setting, as they make the assumption that one can assign arbitrary color to each pixel. In fact, naive application of existing methods fails even in simple domains such as MNIST digits, since one cannot create darker pixels by adding light. We know, however, that the human visual system can be fooled by optical illusions involving certain spatial configurations of brightness and contrast. Our key insight is that one can leverage this behavior to produce high quality images with negligible artifacts. For example, we can create the illusion of darker patches by brightening surrounding pixels. We propose a novel optimization procedure to produce images that satisfy both semantic and non-negativity constraints. Our approach can incorporate existing state-of-the-art methods, and exhibits strong performance in a variety of tasks including image-to-image translation and style transfer.
4.Reducing the Amount of Real World Data for Object Detector Training with Synthetic Data ⬇️
A number of studies have investigated the training of neural networks with synthetic data for applications in the real world. The aim of this study is to quantify how much real world data can be saved when using a mixed dataset of synthetic and real world data. By modeling the relationship between the number of training examples and detection performance by a simple power law, we find that the need for real world data can be reduced by up to 70% without sacrificing detection performance. The training of object detection networks is especially enhanced by enriching the mixed dataset with classes underrepresented in the real world dataset. The results indicate that mixed datasets with real world data ratios between 5% and 20% reduce the need for real world data the most without reducing the detection performance.
5.HCSC: Hierarchical Contrastive Selective Coding ⬇️
Hierarchical semantic structures naturally exist in an image dataset, in which several semantically relevant image clusters can be further integrated into a larger cluster with coarser-grained semantics. Capturing such structures with image representations can greatly benefit the semantic understanding on various downstream tasks. Existing contrastive representation learning methods lack such an important model capability. In addition, the negative pairs used in these methods are not guaranteed to be semantically distinct, which could further hamper the structural correctness of learned image representations. To tackle these limitations, we propose a novel contrastive learning framework called Hierarchical Contrastive Selective Coding (HCSC). In this framework, a set of hierarchical prototypes are constructed and also dynamically updated to represent the hierarchical semantic structures underlying the data in the latent space. To make image representations better fit such semantic structures, we employ and further improve conventional instance-wise and prototypical contrastive learning via an elaborate pair selection scheme. This scheme seeks to select more diverse positive pairs with similar semantics and more precise negative pairs with truly distinct semantics. On extensive downstream tasks, we verify the superior performance of HCSC over state-of-the-art contrastive methods, and the effectiveness of major model components is proved by plentiful analytical studies. Our source code and model weights are available at this https URL
6.Evaluating Feature Attribution: An Information-Theoretic Perspective ⬇️
With a variety of local feature attribution methods being proposed in recent years, follow-up work suggested several evaluation strategies. To assess the attribution quality across different attribution techniques, the most popular among these evaluation strategies in the image domain use pixel perturbations. However, recent advances discovered that different evaluation strategies produce conflicting rankings of attribution methods and can be prohibitively expensive to compute. In this work, we present an information-theoretic analysis of evaluation strategies based on pixel perturbations. Our findings reveal that the results output by different evaluation strategies are strongly affected by information leakage through the shape of the removed pixels as opposed to their actual values. Using our theoretical insights, we propose a novel evaluation framework termed Remove and Debias (ROAD) which offers two contributions: First, it mitigates the impact of the confounders, which entails higher consistency among evaluation strategies. Second, ROAD does not require the computationally expensive retraining step and saves up to 99% in computational costs compared to the state-of-the-art. Our source code is available at this https URL.
7.Sim2Real Object-Centric Keypoint Detection and Description ⬇️
Keypoint detection and description play a central role in computer vision. Most existing methods are in the form of scene-level prediction, without returning the object classes of different keypoints. In this paper, we propose the object-centric formulation, which, beyond the conventional setting, requires further identifying which object each interest point belongs to. With such fine-grained information, our framework enables more downstream potentials, such as object-level matching and pose estimation in a clustered environment. To get around the difficulty of label collection in the real world, we develop a sim2real contrastive learning mechanism that can generalize the model trained in simulation to real-world applications. The novelties of our training method are three-fold: (i) we integrate the uncertainty into the learning framework to improve feature description of hard cases, e.g., less-textured or symmetric patches; (ii) we decouple the object descriptor into two output branches -- intra-object salience and inter-object distinctness, resulting in a better pixel-wise description; (iii) we enforce cross-view semantic consistency for enhanced robustness in representation learning. Comprehensive experiments on image matching and 6D pose estimation verify the encouraging generalization ability of our method from simulation to reality. Particularly for 6D pose estimation, our method significantly outperforms typical unsupervised/sim2real methods, achieving a closer gap with the fully supervised counterpart. Additional results and videos can be found at this https URL
8.Multi-Order Networks for Action Unit Detection ⬇️
Deep multi-task methods, where several tasks are learned within a single network, have recently attracted increasing attention. Burning point of this attention is their capacity to capture inter-task relationships. Current approaches either only rely on weight sharing, or add explicit dependency modelling by decomposing the task joint distribution using Bayes chain rule. If the latter strategy yields comprehensive inter-task relationships modelling, it requires imposing an arbitrary order into an unordered task set. Most importantly, this sequence ordering choice has been identified as a critical source of performance variations. In this paper, we present Multi-Order Network (MONET), a multi-task learning method with joint task order optimization. MONET uses a differentiable order selection based on soft order modelling inside Birkhoff's polytope to jointly learn task-wise recurrent modules with their optimal chaining order. Furthermore, we introduce warm up and order dropout to enhance order selection by encouraging order exploration. Experimentally, we first validate MONET capacity to retrieve the optimal order in a toy environment. Second, we use an attribute detection scenario to show that MONET outperforms existing multi-task baselines on a wide range of dependency settings. Finally, we demonstrate that MONET significantly extends state-of-the-art performance in Facial Action Unit detection.
9.Continual Attentive Fusion for Incremental Learning in Semantic Segmentation ⬇️
Over the past years, semantic segmentation, as many other tasks in computer vision, benefited from the progress in deep neural networks, resulting in significantly improved performance. However, deep architectures trained with gradient-based techniques suffer from catastrophic forgetting, which is the tendency to forget previously learned knowledge while learning new tasks. Aiming at devising strategies to counteract this effect, incremental learning approaches have gained popularity over the past years. However, the first incremental learning methods for semantic segmentation appeared only recently. While effective, these approaches do not account for a crucial aspect in pixel-level dense prediction problems, i.e. the role of attention mechanisms. To fill this gap, in this paper we introduce a novel attentive feature distillation approach to mitigate catastrophic forgetting while accounting for semantic spatial- and channel-level dependencies. Furthermore, we propose a {continual attentive fusion} structure, which takes advantage of the attention learned from the new and the old tasks while learning features for the new task. Finally, we also introduce a novel strategy to account for the background class in the distillation loss, thus preventing biased predictions. We demonstrate the effectiveness of our approach with an extensive evaluation on Pascal-VOC 2012 and ADE20K, setting a new state of the art.
10.Review of Serial and Parallel Min-Cut/Max-Flow Algorithms for Computer Vision ⬇️
Minimum cut / maximum flow (min-cut/max-flow) algorithms are used to solve a variety of problems in computer vision and thus significant effort has been put into developing fast min-cut/max-flow algorithms. This makes it difficult to choose an optimal algorithm for a given problem - especially for parallel algorithms, which have not been thoroughly compared. In this paper, we review the state-of-the-art min-cut/max-flow algorithms for unstructured graphs in computer vision. We evaluate run time performance and memory use of various implementations of both serial and parallel algorithms on a set of graph cut problems. Our results show that the Hochbaum pseudoflow algorithm is the fastest serial algorithm closely followed by the Excesses Incremental Breadth First Search algorithm, while the Boykov-Kolmogorov algorithm is the most memory efficient. The best parallel algorithm is the adaptive bottom-up merging approach by Liu and Sun. Additionally, we show significant variations in performance between different implementations the same algorithms highlighting the importance of low-level implementation details. Finally, we note that existing parallel min-cut/max-flow algorithms can significantly outperform serial algorithms on large problems but suffers from added overhead on small to medium problems. Implementations of all algorithms are available at this https URL
11.Filtered-CoPhy: Unsupervised Learning of Counterfactual Physics in Pixel Space ⬇️
Learning causal relationships in high-dimensional data (images, videos) is a hard task, as they are often defined on low dimensional manifolds and must be extracted from complex signals dominated by appearance, lighting, textures and also spurious correlations in the data. We present a method for learning counterfactual reasoning of physical processes in pixel space, which requires the prediction of the impact of interventions on initial conditions. Going beyond the identification of structural relationships, we deal with the challenging problem of forecasting raw video over long horizons. Our method does not require the knowledge or supervision of any ground truth positions or other object or scene properties. Our model learns and acts on a suitable hybrid latent representation based on a combination of dense features, sets of 2D keypoints and an additional latent vector per keypoint. We show that this better captures the dynamics of physical processes than purely dense or sparse representations. We introduce a new challenging and carefully designed counterfactual benchmark for predictions in pixel space and outperform strong baselines in physics-inspired ML and video prediction.
12.From Explanations to Segmentation: Using Explainable AI for Image Segmentation ⬇️
The new era of image segmentation leveraging the power of Deep Neural Nets (DNNs) comes with a price tag: to train a neural network for pixel-wise segmentation, a large amount of training samples has to be manually labeled on pixel-precision. In this work, we address this by following an indirect solution. We build upon the advances of the Explainable AI (XAI) community and extract a pixel-wise binary segmentation from the output of the Layer-wise Relevance Propagation (LRP) explaining the decision of a classification network. We show that we achieve similar results compared to an established U-Net segmentation architecture, while the generation of the training data is significantly simplified. The proposed method can be trained in a weakly supervised fashion, as the training samples must be only labeled on image-level, at the same time enabling the output of a segmentation mask. This makes it especially applicable to a wider range of real applications where tedious pixel-level labelling is often not possible.
13.Laplacian2Mesh: Laplacian-Based Mesh Understanding ⬇️
Geometric deep learning has sparked a rising interest in computer graphics to perform shape understanding tasks, such as shape classification and semantic segmentation on three-dimensional (3D) geometric surfaces. Previous works explored the significant direction by defining the operations of convolution and pooling on triangle meshes, but most methods explicitly utilized the graph connection structure of the mesh. Motivated by the geometric spectral surface reconstruction theory, we introduce a novel and flexible convolutional neural network (CNN) model, called Laplacian2Mesh, for 3D triangle mesh, which maps the features of mesh in the Euclidean space to the multi-dimensional Laplacian-Beltrami space, which is similar to the multi-resolution input in 2D CNN. Mesh pooling is applied to expand the receptive field of the network by the multi-space transformation of Laplacian which retains the surface topology, and channel self-attention convolutions are applied in the new space. Since implicitly using the intrinsic geodesic connections of the mesh through the adjacency matrix, we do not consider the number of the neighbors of the vertices, thereby mesh data with different numbers of vertices can be input. Experiments on various learning tasks applied to 3D meshes demonstrate the effectiveness and efficiency of Laplacian2Mesh.
14.Access Control of Object Detection Models Using Encrypted Feature Maps ⬇️
In this paper, we propose an access control method for object detection models. The use of encrypted images or encrypted feature maps has been demonstrated to be effective in access control of models from unauthorized access. However, the effectiveness of the approach has been confirmed in only image classification models and semantic segmentation models, but not in object detection models. In this paper, the use of encrypted feature maps is shown to be effective in access control of object detection models for the first time.
15.Detecting Human-Object Interactions with Object-Guided Cross-Modal Calibrated Semantics ⬇️
Human-Object Interaction (HOI) detection is an essential task to understand human-centric images from a fine-grained perspective. Although end-to-end HOI detection models thrive, their paradigm of parallel human/object detection and verb class prediction loses two-stage methods' merit: object-guided hierarchy. The object in one HOI triplet gives direct clues to the verb to be predicted. In this paper, we aim to boost end-to-end models with object-guided statistical priors. Specifically, We propose to utilize a Verb Semantic Model (VSM) and use semantic aggregation to profit from this object-guided hierarchy. Similarity KL (SKL) loss is proposed to optimize VSM to align with the HOI dataset's priors. To overcome the static semantic embedding problem, we propose to generate cross-modality-aware visual and semantic features by Cross-Modal Calibration (CMC). The above modules combined composes Object-guided Cross-modal Calibration Network (OCN). Experiments conducted on two popular HOI detection benchmarks demonstrate the significance of incorporating the statistical prior knowledge and produce state-of-the-art performances. More detailed analysis indicates proposed modules serve as a stronger verb predictor and a more superior method of utilizing prior knowledge. The codes are available at \url{this https URL}.
16.Semi-supervised 3D Object Detection via Temporal Graph Neural Networks ⬇️
3D object detection plays an important role in autonomous driving and other robotics applications. However, these detectors usually require training on large amounts of annotated data that is expensive and time-consuming to collect. Instead, we propose leveraging large amounts of unlabeled point cloud videos by semi-supervised learning of 3D object detectors via temporal graph neural networks. Our insight is that temporal smoothing can create more accurate detection results on unlabeled data, and these smoothed detections can then be used to retrain the detector. We learn to perform this temporal reasoning with a graph neural network, where edges represent the relationship between candidate detections in different time frames. After semi-supervised learning, our method achieves state-of-the-art detection performance on the challenging nuScenes and H3D benchmarks, compared to baselines trained on the same amount of labeled data. Project and code are released at this https URL.
17.CLA-NeRF: Category-Level Articulated Neural Radiance Field ⬇️
We propose CLA-NeRF -- a Category-Level Articulated Neural Radiance Field that can perform view synthesis, part segmentation, and articulated pose estimation. CLA-NeRF is trained at the object category level using no CAD models and no depth, but a set of RGB images with ground truth camera poses and part segments. During inference, it only takes a few RGB views (i.e., few-shot) of an unseen 3D object instance within the known category to infer the object part segmentation and the neural radiance field. Given an articulated pose as input, CLA-NeRF can perform articulation-aware volume rendering to generate the corresponding RGB image at any camera pose. Moreover, the articulated pose of an object can be estimated via inverse rendering. In our experiments, we evaluate the framework across five categories on both synthetic and real-world data. In all cases, our method shows realistic deformation results and accurate articulated pose estimation. We believe that both few-shot articulated object rendering and articulated pose estimation open doors for robots to perceive and interact with unseen articulated objects.
18.Dilated Continuous Random Field for Semantic Segmentation ⬇️
Mean field approximation methodology has laid the foundation of modern Continuous Random Field (CRF) based solutions for the refinement of semantic segmentation. In this paper, we propose to relax the hard constraint of mean field approximation - minimizing the energy term of each node from probabilistic graphical model, by a global optimization with the proposed dilated sparse convolution module (DSConv). In addition, adaptive global average-pooling and adaptive global max-pooling are implemented as replacements of fully connected layers. In order to integrate DSConv, we design an end-to-end, time-efficient DilatedCRF pipeline. The unary energy term is derived either from pre-softmax and post-softmax features, or the predicted affordance map using a conventional classifier, making it easier to implement DilatedCRF for varieties of classifiers. We also present superior experimental results of proposed approach on the suction dataset comparing to other CRF-based approaches.
19.Learning-Based Framework for Camera Calibration with Distortion Correction and High Precision Feature Detection ⬇️
Camera calibration is a crucial technique which significantly influences the performance of many robotic systems. Robustness and high precision have always been the pursuit of diverse calibration methods. State-of-the-art calibration techniques based on classical Zhang's method, however, still suffer from environmental noise, radial lens distortion and sub-optimal parameter estimation. Therefore, in this paper, we propose a hybrid camera calibration framework which combines learning-based approaches with traditional methods to handle these bottlenecks. In particular, this framework leverages learning-based approaches to perform efficient distortion correction and robust chessboard corner coordinate encoding. For sub-pixel accuracy of corner detection, a specially-designed coordinate decoding algorithm with embed outlier rejection mechanism is proposed. To avoid sub-optimal estimation results, we improve the traditional parameter estimation by RANSAC algorithm and achieve stable results. Compared with two widely-used camera calibration toolboxes, experiment results on both real and synthetic datasets manifest the better robustness and higher precision of the proposed framework. The massive synthetic dataset is the basis of our framework's decent performance and will be publicly available along with the code at this https URL.
20.3D Visualization and Spatial Data Mining for Analysis of LULC Images ⬇️
The present study is an attempt made to create a new tool for the analysis of Land Use Land Cover (LUCL) images in 3D visualization. This study mainly uses spatial data mining techniques on high resolution LULC satellite imagery. Visualization of feature space allows exploration of patterns in the image data and insight into the classification process and related uncertainty. Visual Data Mining provides added value to image classifications as the user can be involved in the classification process providing increased confidence in and understanding of the results. In this study, we present a prototype of image segmentation, K-Means clustering and 3D visualization tool for visual data mining (VDM) of LUCL satellite imagery into volume visualization. This volume based representation divides feature space into spheres or voxels. The visualization tool is showcased in a classification study of high-resolution LULC imagery of Latur district (Maharashtra state, India) is used as sample data.
21.Real-Time Facial Expression Recognition using Facial Landmarks and Neural Networks ⬇️
This paper presents a lightweight algorithm for feature extraction, classification of seven different emotions, and facial expression recognition in a real-time manner based on static images of the human face. In this regard, a Multi-Layer Perceptron (MLP) neural network is trained based on the foregoing algorithm. In order to classify human faces, first, some pre-processing is applied to the input image, which can localize and cut out faces from it. In the next step, a facial landmark detection library is used, which can detect the landmarks of each face. Then, the human face is split into upper and lower faces, which enables the extraction of the desired features from each part. In the proposed model, both geometric and texture-based feature types are taken into account. After the feature extraction phase, a normalized vector of features is created. A 3-layer MLP is trained using these feature vectors, leading to 96% accuracy on the test set.
22.Semi-supervised Identification and Mapping of Surface Water Extent using Street-level Monitoring Videos ⬇️
Urban flooding is becoming a common and devastating hazard to cause life loss and economic damage. Monitoring and understanding urban flooding in the local scale is a challenging task due to the complicated urban landscape, intricate hydraulic process, and the lack of high-quality and resolution data. The emerging smart city technology such as monitoring cameras provides an unprecedented opportunity to address the data issue. However, estimating the water accumulation on the land surface based on the monitoring footage is unreliable using the traditional segmentation technique because the boundary of the water accumulation, under the influence of varying weather, background, and illumination, is usually too fuzzy to identify, and the oblique angle and image distortion in the video monitoring data prevents georeferencing and object-based measurements. This paper presents a novel semi-supervised segmentation scheme for surface water extent recognition from the footage of an oblique monitoring camera. The semi-supervised segmentation algorithm was found suitable to determine the water boundary and the monoplotting method was successfully applied to georeference the pixels of the monitoring video for the virtual quantification of the local drainage process. The correlation and mechanism-based analysis demonstrates the value of the proposed method in advancing the understanding of local drainage hydraulics. The workflow and created methods in this study has a great potential to study other street-level and earth surface processes.
23.Deep-Disaster: Unsupervised Disaster Detection and Localization Using Visual Data ⬇️
Social media plays a significant role in sharing essential information, which helps humanitarian organizations in rescue operations during and after disaster incidents. However, developing an efficient method that can provide rapid analysis of social media images in the early hours of disasters is still largely an open problem, mainly due to the lack of suitable datasets and the sheer complexity of this task. In addition, supervised methods can not generalize well to novel disaster incidents. In this paper, inspired by the success of Knowledge Distillation (KD) methods, we propose an unsupervised deep neural network to detect and localize damages in social media images. Our proposed KD architecture is a feature-based distillation approach that comprises a pre-trained teacher and a smaller student network, with both networks having similar GAN architecture containing a generator and a discriminator. The student network is trained to emulate the behavior of the teacher on training input samples, which, in turn, contain images that do not include any damaged regions. Therefore, the student network only learns the distribution of no damage data and would have different behavior from the teacher network-facing damages. To detect damage, we utilize the difference between features generated by two networks using a defined score function that demonstrates the probability of damages occurring. Our experimental results on the benchmark dataset confirm that our approach outperforms state-of-the-art methods in detecting and localizing the damaged areas, especially for novel disaster types.
24.Finding Directions in GAN's Latent Space for Neural Face Reenactment ⬇️
This paper is on face/head reenactment where the goal is to transfer the facial pose (3D head orientation and expression) of a target face to a source face. Previous methods focus on learning embedding networks for identity and pose disentanglement which proves to be a rather hard task, degrading the quality of the generated images. We take a different approach, bypassing the training of such networks, by using (fine-tuned) pre-trained GANs which have been shown capable of producing high-quality facial images. Because GANs are characterized by weak controllability, the core of our approach is a method to discover which directions in latent GAN space are responsible for controlling facial pose and expression variations. We present a simple pipeline to learn such directions with the aid of a 3D shape model which, by construction, already captures disentangled directions for facial pose, identity and expression. Moreover, we show that by embedding real images in the GAN latent space, our method can be successfully used for the reenactment of real-world faces. Our method features several favorable properties including using a single source image (one-shot) and enabling cross-person reenactment. Our qualitative and quantitative results show that our approach often produces reenacted faces of significantly higher quality than those produced by state-of-the-art methods for the standard benchmarks of VoxCeleb1 & 2.
25.AI-based Medical e-Diagnosis for Fast and Automatic Ventricular Volume Measurement in the Patients with Normal Pressure Hydrocephalus ⬇️
Based on CT and MRI images acquired from normal pressure hydrocephalus (NPH) patients, using machine learning methods, we aim to establish a multi-modal and high-performance automatic ventricle segmentation method to achieve efficient and accurate automatic measurement of the ventricular volume. First, we extract the brain CT and MRI images of 143 definite NPH patients. Second, we manually label the ventricular volume (VV) and intracranial volume (ICV). Then, we use machine learning method to extract features and establish automatic ventricle segmentation model. Finally, we verify the reliability of the model and achieved automatic measurement of VV and ICV. In CT images, the Dice similarity coefficient (DSC), Intraclass Correlation Coefficient (ICC), Pearson correlation, and Bland-Altman analysis of the automatic and manual segmentation result of the VV were 0.95, 0.99, 0.99, and 4.2$\pm$2.6 respectively. The results of ICV were 0.96, 0.99, 0.99, and 6.0$\pm$3.8 respectively. The whole process takes 3.4$\pm$0.3 seconds. In MRI images, the DSC, ICC, Pearson correlation, and Bland-Altman analysis of the automatic and manual segmentation result of the VV were 0.94, 0.99, 0.99, and 2.0$\pm$0.6 respectively. The results of ICV were 0.93, 0.99, 0.99, and 7.9$\pm$3.8 respectively. The whole process took 1.9$\pm$0.1 seconds. We have established a multi-modal and high-performance automatic ventricle segmentation method to achieve efficient and accurate automatic measurement of the ventricular volume of NPH patients. This can help clinicians quickly and accurately understand the situation of NPH patient's ventricles.
26.Datamodels: Predicting Predictions from Training Data ⬇️
We present a conceptual framework, datamodeling, for analyzing the behavior of a model class in terms of the training data. For any fixed "target" example
$x$ , training set$S$ , and learning algorithm, a datamodel is a parameterized function$2^S \to \mathbb{R}$ that for any subset of$S' \subset S$ -- using only information about which examples of$S$ are contained in$S'$ -- predicts the outcome of training a model on$S'$ and evaluating on$x$ . Despite the potential complexity of the underlying process being approximated (e.g., end-to-end training and evaluation of deep neural networks), we show that even simple linear datamodels can successfully predict model outputs. We then demonstrate that datamodels give rise to a variety of applications, such as: accurately predicting the effect of dataset counterfactuals; identifying brittle predictions; finding semantically similar examples; quantifying train-test leakage; and embedding data into a well-behaved and feature-rich representation space. Data for this paper (including pre-computed datamodels as well as raw predictions from four million trained deep neural networks) is available at this https URL .
27.Fishing for User Data in Large-Batch Federated Learning via Gradient Magnification ⬇️
Federated learning (FL) has rapidly risen in popularity due to its promise of privacy and efficiency. Previous works have exposed privacy vulnerabilities in the FL pipeline by recovering user data from gradient updates. However, existing attacks fail to address realistic settings because they either 1) require a `toy' settings with very small batch sizes, or 2) require unrealistic and conspicuous architecture modifications. We introduce a new strategy that dramatically elevates existing attacks to operate on batches of arbitrarily large size, and without architectural modifications. Our model-agnostic strategy only requires modifications to the model parameters sent to the user, which is a realistic threat model in many scenarios. We demonstrate the strategy in challenging large-scale settings, obtaining high-fidelity data extraction in both cross-device and cross-silo federated learning.
28.Development of a neural network to recognize standards and features from 3D CAD models ⬇️
Focus of this work is to recognize standards and further features directly from 3D CAD models. For this reason, a neural network was trained to recognize nine classes of machine elements. After the system identified a part as a standard, like a hexagon head screw after the DIN EN ISO 8676, it accesses the geometrical information of the CAD system via the Application Programming Interface (API). In the API, the system searches for necessary information to describe the part appropriately. Based on this information standardized parts can be recognized in detail and supplemented with further information.
29.The impact of removing head movements on audio-visual speech enhancement ⬇️
This paper investigates the impact of head movements on audio-visual speech enhancement (AVSE). Although being a common conversational feature, head movements have been ignored by past and recent studies: they challenge today's learning-based methods as they often degrade the performance of models that are trained on clean, frontal, and steady face images. To alleviate this problem, we propose to use robust face frontalization (RFF) in combination with an AVSE method based on a variational auto-encoder (VAE) model. We briefly describe the basic ingredients of the proposed pipeline and we perform experiments with a recently released audio-visual dataset. In the light of these experiments, and based on three standard metrics, namely STOI, PESQ and SI-SDR, we conclude that RFF improves the performance of AVSE by a considerable margin.
30.A generalizable approach based on U-Net model for automatic Intra retinal cyst segmentation in SD-OCT images ⬇️
Intra retinal fluids or Cysts are one of the important symptoms of macular pathologies that are efficiently visualized in OCT images. Automatic segmentation of these abnormalities has been widely investigated in medical image processing studies. In this paper, we propose a new U-Net-based approach for Intra retinal cyst segmentation across different vendors that improves some of the challenges faced by previous deep-based techniques. The proposed method has two main steps: 1- prior information embedding and input data adjustment, and 2- IRC segmentation model. In the first step, we inject the information into the network in a way that overcomes some of the network limitations in receiving data and learning important contextual knowledge. And in the next step, we introduced a connection module between encoder and decoder parts of the standard U-Net architecture that transfers information more effectively from the encoder to the decoder part. Two public datasets namely OPTIMA and KERMANY were employed to evaluate the proposed method. Results showed that the proposed method is an efficient vendor-independent approach for IRC segmentation with mean Dice values of 0.78 and 0.81 on the OPTIMA and KERMANY datasets, respectively.
31.Sinogram Enhancement with Generative Adversarial Networks using Shape Priors ⬇️
Compensating scarce measurements by inferring them from computational models is a way to address ill-posed inverse problems. We tackle Limited Angle Tomography by completing the set of acquisitions using a generative model and prior-knowledge about the scanned object. Using a Generative Adversarial Network as model and Computer-Assisted Design data as shape prior, we demonstrate a quantitative and qualitative advantage of our technique over other state-of-the-art methods. Inferring a substantial number of consecutive missing measurements, we offer an alternative to other image inpainting techniques that fall short of providing a satisfying answer to our research question: can X-Ray exposition be reduced by using generative models to infer lacking measurements?
32.CAESR: Conditional Autoencoder and Super-Resolution for Learned Spatial Scalability ⬇️
In this paper, we present CAESR, an hybrid learning-based coding approach for spatial scalability based on the versatile video coding (VVC) standard. Our framework considers a low-resolution signal encoded with VVC intra-mode as a base-layer (BL), and a deep conditional autoencoder with hyperprior (AE-HP) as an enhancement-layer (EL) model. The EL encoder takes as inputs both the upscaled BL reconstruction and the original image. Our approach relies on conditional coding that learns the optimal mixture of the source and the upscaled BL image, enabling better performance than residual coding. On the decoder side, a super-resolution (SR) module is used to recover high-resolution details and invert the conditional coding process. Experimental results have shown that our solution is competitive with the VVC full-resolution intra coding while being scalable.
33.Minority Class Oriented Active Learning for Imbalanced Datasets ⬇️
Active learning aims to optimize the dataset annotation process when resources are constrained. Most existing methods are designed for balanced datasets. Their practical applicability is limited by the fact that a majority of real-life datasets are actually imbalanced. Here, we introduce a new active learning method which is designed for imbalanced datasets. It favors samples likely to be in minority classes so as to reduce the imbalance of the labeled subset and create a better representation for these classes. We also compare two training schemes for active learning: (1) the one commonly deployed in deep active learning using model fine tuning for each iteration and (2) a scheme which is inspired by transfer learning and exploits generic pre-trained models and train shallow classifiers for each iteration. Evaluation is run with three imbalanced datasets. Results show that the proposed active learning method outperforms competitive baselines. Equally interesting, they also indicate that the transfer learning training scheme outperforms model fine tuning if features are transferable from the generic dataset to the unlabeled one. This last result is surprising and should encourage the community to explore the design of deep active learning methods.
34.A Comparative Study of Calibration Methods for Imbalanced Class Incremental Learning ⬇️
Deep learning approaches are successful in a wide range of AI problems and in particular for visual recognition tasks. However, there are still open problems among which is the capacity to handle streams of visual information and the management of class imbalance in datasets. Existing research approaches these two problems separately while they co-occur in real world applications. Here, we study the problem of learning incrementally from imbalanced datasets. We focus on algorithms which have a constant deep model complexity and use a bounded memory to store exemplars of old classes across incremental states. Since memory is bounded, old classes are learned with fewer images than new classes and an imbalance due to incremental learning is added to the initial dataset imbalance. A score prediction bias in favor of new classes appears and we evaluate a comprehensive set of score calibration methods to reduce it. Evaluation is carried with three datasets, using two dataset imbalance configurations and three bounded memory sizes. Results show that most calibration methods have beneficial effect and that they are most useful for lower bounded memory sizes, which are most interesting in practice. As a secondary contribution, we remove the usual distillation component from the loss function of incremental learning algorithms. We show that simpler vanilla fine tuning is a stronger backbone for imbalanced incremental learning algorithms.
35.StyleGAN-XL: Scaling StyleGAN to Large Diverse Datasets ⬇️
Computer graphics has experienced a recent surge of data-centric approaches for photorealistic and controllable content creation. StyleGAN in particular sets new standards for generative modeling regarding image quality and controllability. However, StyleGAN's performance severely degrades on large unstructured datasets such as ImageNet. StyleGAN was designed for controllability; hence, prior works suspect its restrictive design to be unsuitable for diverse datasets. In contrast, we find the main limiting factor to be the current training strategy. Following the recently introduced Projected GAN paradigm, we leverage powerful neural network priors and a progressive growing strategy to successfully train the latest StyleGAN3 generator on ImageNet. Our final model, StyleGAN-XL, sets a new state-of-the-art on large-scale image synthesis and is the first to generate images at a resolution of
$1024^2$ at such a dataset scale. We demonstrate that this model can invert and edit images beyond the narrow domain of portraits or specific object classes.
36.Fully Online Meta-Learning Without Task Boundaries ⬇️
While deep networks can learn complex functions such as classifiers, detectors, and trackers, many applications require models that continually adapt to changing input distributions, changing tasks, and changing environmental conditions. Indeed, this ability to continuously accrue knowledge and use past experience to learn new tasks quickly in continual settings is one of the key properties of an intelligent system. For complex and high-dimensional problems, simply updating the model continually with standard learning algorithms such as gradient descent may result in slow adaptation. Meta-learning can provide a powerful tool to accelerate adaptation yet is conventionally studied in batch settings. In this paper, we study how meta-learning can be applied to tackle online problems of this nature, simultaneously adapting to changing tasks and input distributions and meta-training the model in order to adapt more quickly in the future. Extending meta-learning into the online setting presents its own challenges, and although several prior methods have studied related problems, they generally require a discrete notion of tasks, with known ground-truth task boundaries. Such methods typically adapt to each task in sequence, resetting the model between tasks, rather than adapting continuously across tasks. In many real-world settings, such discrete boundaries are unavailable, and may not even exist. To address these settings, we propose a Fully Online Meta-Learning (FOML) algorithm, which does not require any ground truth knowledge about the task boundaries and stays fully online without resetting back to pre-trained weights. Our experiments show that FOML was able to learn new tasks faster than the state-of-the-art online learning methods on Rainbow-MNIST, CIFAR100 and CELEBA datasets.
37.Adversarial Imitation Learning from Video using a State Observer ⬇️
The imitation learning research community has recently made significant progress towards the goal of enabling artificial agents to imitate behaviors from video demonstrations alone. However, current state-of-the-art approaches developed for this problem exhibit high sample complexity due, in part, to the high-dimensional nature of video observations. Towards addressing this issue, we introduce here a new algorithm called Visual Generative Adversarial Imitation from Observation using a State Observer VGAIfO-SO. At its core, VGAIfO-SO seeks to address sample inefficiency using a novel, self-supervised state observer, which provides estimates of lower-dimensional proprioceptive state representations from high-dimensional images. We show experimentally in several continuous control environments that VGAIfO-SO is more sample efficient than other IfO algorithms at learning from video-only demonstrations and can sometimes even achieve performance close to the Generative Adversarial Imitation from Observation (GAIfO) algorithm that has privileged access to the demonstrator's proprioceptive state information.
38.ISNet: Costless and Implicit Image Segmentation for Deep Classifiers, with Application in COVID-19 Detection ⬇️
In this work we propose a novel deep neural network (DNN) architecture, ISNet, to solve the task of image segmentation followed by classification, substituting the common pipeline of two networks by a single model. We designed the ISNet for high flexibility and performance: it allows virtually any classification neural network architecture to analyze a common image as if it had been previously segmented. Furthermore, in relation to the original classifier, the ISNet does not cause any increment in computational cost or architectural changes at run-time. To accomplish this, we introduce the concept of optimizing DNNs for relevance segmentation in heatmaps created by Layer-wise Relevance Propagation (LRP), which proves to be equivalent to the classification of previously segmented images. We apply an ISNet based on a DenseNet121 classifier to solve the task of COVID-19 detection in chest X-rays. We compare the model to a U-net (performing lung segmentation) followed by a DenseNet121, and to a standalone DenseNet121. Due to the implicit segmentation, the ISNet precisely ignored the X-ray regions outside of the lungs; it achieved 94.5 +/-4.1% mean accuracy with an external database, showing strong generalization capability and surpassing the other models' performances by 6 to 7.9%. ISNet presents a fast and light methodology to perform classification preceded by segmentation, while also being more accurate than standard pipelines.
39.Disentangling multiple scattering with deep learning: application to strain mapping from electron diffraction patterns ⬇️
Implementation of a fast, robust, and fully-automated pipeline for crystal structure determination and underlying strain mapping for crystalline materials is important for many technological applications. Scanning electron nanodiffraction offers a procedure for identifying and collecting strain maps with good accuracy and high spatial resolutions. However, the application of this technique is limited, particularly in thick samples where the electron beam can undergo multiple scattering, which introduces signal nonlinearities. Deep learning methods have the potential to invert these complex signals, but previous implementations are often trained only on specific crystal systems or a small subset of the crystal structure and microscope parameter phase space. In this study, we implement a Fourier space, complex-valued deep neural network called FCU-Net, to invert highly nonlinear electron diffraction patterns into the corresponding quantitative structure factor images. We trained the FCU-Net using over 200,000 unique simulated dynamical diffraction patterns which include many different combinations of crystal structures, orientations, thicknesses, microscope parameters, and common experimental artifacts. We evaluated the trained FCU-Net model against simulated and experimental 4D-STEM diffraction datasets, where it substantially out-performs conventional analysis methods. Our simulated diffraction pattern library, implementation of FCU-Net, and trained model weights are freely available in open source repositories, and can be adapted to many different diffraction measurement problems.
40.Recognition-Aware Learned Image Compression ⬇️
Learned image compression methods generally optimize a rate-distortion loss, trading off improvements in visual distortion for added bitrate. Increasingly, however, compressed imagery is used as an input to deep learning networks for various tasks such as classification, object detection, and superresolution. We propose a recognition-aware learned compression method, which optimizes a rate-distortion loss alongside a task-specific loss, jointly learning compression and recognition networks. We augment a hierarchical autoencoder-based compression network with an EfficientNet recognition model and use two hyperparameters to trade off between distortion, bitrate, and recognition performance. We characterize the classification accuracy of our proposed method as a function of bitrate and find that for low bitrates our method achieves as much as 26% higher recognition accuracy at equivalent bitrates compared to traditional methods such as Better Portable Graphics (BPG).
41.ATEK: Augmenting Transformers with Expert Knowledge for Indoor Layout Synthesis ⬇️
We address the problem of indoor layout synthesis, which is a topic of continuing research interest in computer graphics. The newest works made significant progress using data-driven generative methods; however, these approaches rely on suitable datasets. In practice, desirable layout properties may not exist in a dataset, for instance, specific expert knowledge can be missing in the data. We propose a method that combines expert knowledge, for example, knowledge about ergonomics, with a data-driven generator based on the popular Transformer architecture. The knowledge is given as differentiable scalar functions, which can be used both as weights or as additional terms in the loss function. Using this knowledge, the synthesized layouts can be biased to exhibit desirable properties, even if these properties are not present in the dataset. Our approach can also alleviate problems of lack of data and imperfections in the data. Our work aims to improve generative machine learning for modeling and provide novel tools for designers and amateurs for the problem of interior layout creation.
42.Blind Image Deconvolution Using Variational Deep Image Prior ⬇️
Conventional deconvolution methods utilize hand-crafted image priors to constrain the optimization. While deep-learning-based methods have simplified the optimization by end-to-end training, they fail to generalize well to blurs unseen in the training dataset. Thus, training image-specific models is important for higher generalization. Deep image prior (DIP) provides an approach to optimize the weights of a randomly initialized network with a single degraded image by maximum a posteriori (MAP), which shows that the architecture of a network can serve as the hand-crafted image prior. Different from the conventional hand-crafted image priors that are statistically obtained, it is hard to find a proper network architecture because the relationship between images and their corresponding network architectures is unclear. As a result, the network architecture cannot provide enough constraint for the latent sharp image. This paper proposes a new variational deep image prior (VDIP) for blind image deconvolution, which exploits additive hand-crafted image priors on latent sharp images and approximates a distribution for each pixel to avoid suboptimal solutions. Our mathematical analysis shows that the proposed method can better constrain the optimization. The experimental results further demonstrate that the generated images have better quality than that of the original DIP on benchmark datasets. The source code of our VDIP is available at this https URL.
43.Fractional Motion Estimation for Point Cloud Compression ⬇️
Motivated by the success of fractional pixel motion in video coding, we explore the design of motion estimation with fractional-voxel resolution for compression of color attributes of dynamic 3D point clouds. Our proposed block-based fractional-voxel motion estimation scheme takes into account the fundamental differences between point clouds and videos, i.e., the irregularity of the distribution of voxels within a frame and across frames. We show that motion compensation can benefit from the higher resolution reference and more accurate displacements provided by fractional precision. Our proposed scheme significantly outperforms comparable methods that only use integer motion. The proposed scheme can be combined with and add sizeable gains to state-of-the-art systems that use transforms such as Region Adaptive Graph Fourier Transform and Region Adaptive Haar Transform.
44.DexVIP: Learning Dexterous Grasping with Human Hand Pose Priors from Video ⬇️
Dexterous multi-fingered robotic hands have a formidable action space, yet their morphological similarity to the human hand holds immense potential to accelerate robot learning. We propose DexVIP, an approach to learn dexterous robotic grasping from human-object interactions present in in-the-wild YouTube videos. We do this by curating grasp images from human-object interaction videos and imposing a prior over the agent's hand pose when learning to grasp with deep reinforcement learning. A key advantage of our method is that the learned policy is able to leverage free-form in-the-wild visual data. As a result, it can easily scale to new objects, and it sidesteps the standard practice of collecting human demonstrations in a lab -- a much more expensive and indirect way to capture human expertise. Through experiments on 27 objects with a 30-DoF simulated robot hand, we demonstrate that DexVIP compares favorably to existing approaches that lack a hand pose prior or rely on specialized tele-operation equipment to obtain human demonstrations, while also being faster to train. Project page: this https URL
45.Query Efficient Decision Based Sparse Attacks Against Black-Box Deep Learning Models ⬇️
Despite our best efforts, deep learning models remain highly vulnerable to even tiny adversarial perturbations applied to the inputs. The ability to extract information from solely the output of a machine learning model to craft adversarial perturbations to black-box models is a practical threat against real-world systems, such as autonomous cars or machine learning models exposed as a service (MLaaS). Of particular interest are sparse attacks. The realization of sparse attacks in black-box models demonstrates that machine learning models are more vulnerable than we believe. Because these attacks aim to minimize the number of perturbed pixels measured by l_0 norm-required to mislead a model by solely observing the decision (the predicted label) returned to a model query; the so-called decision-based attack setting. But, such an attack leads to an NP-hard optimization problem. We develop an evolution-based algorithm-SparseEvo-for the problem and evaluate against both convolutional deep neural networks and vision transformers. Notably, vision transformers are yet to be investigated under a decision-based attack setting. SparseEvo requires significantly fewer model queries than the state-of-the-art sparse attack Pointwise for both untargeted and targeted attacks. The attack algorithm, although conceptually simple, is also competitive with only a limited query budget against the state-of-the-art gradient-based whitebox attacks in standard computer vision tasks such as ImageNet. Importantly, the query efficient SparseEvo, along with decision-based attacks, in general, raise new questions regarding the safety of deployed systems and poses new directions to study and understand the robustness of machine learning models.
46.Holistic Fine-grained GGS Characterization: From Detection to Unbalanced Classification ⬇️
Recent studies have demonstrated the diagnostic and prognostic values of global glomerulosclerosis (GGS) in IgA nephropathy, aging, and end-stage renal disease. However, the fine-grained quantitative analysis of multiple GGS subtypes (e.g., obsolescent, solidified, and disappearing glomerulosclerosis) is typically a resource extensive manual process. Very few automatic methods, if any, have been developed to bridge this gap for such analytics. In this paper, we present a holistic pipeline to quantify GGS (with both detection and classification) from a whole slide image in a fully automatic manner. In addition, we conduct the fine-grained classification for the sub-types of GGS. Our study releases the open-source quantitative analytical tool for fine-grained GGS characterization while tackling the technical challenges in unbalanced classification and integrating detection and classification.
47.AutoGeoLabel: Automated Label Generation for Geospatial Machine Learning ⬇️
A key challenge of supervised learning is the availability of human-labeled data. We evaluate a big data processing pipeline to auto-generate labels for remote sensing data. It is based on rasterized statistical features extracted from surveys such as e.g. LiDAR measurements. Using simple combinations of the rasterized statistical layers, it is demonstrated that multiple classes can be generated at accuracies of ~0.9. As proof of concept, we utilize the big geo-data platform IBM PAIRS to dynamically generate such labels in dense urban areas with multiple land cover classes. The general method proposed here is platform independent, and it can be adapted to generate labels for other satellite modalities in order to enable machine learning on overhead imagery for land use classification and object detection.
48.Leveraging Bitstream Metadata for Fast and Accurate Video Compression Correction ⬇️
Video compression is a central feature of the modern internet powering technologies from social media to video conferencing. While video compression continues to mature, for many, and particularly for extreme, compression settings, quality loss is still noticeable. These extreme settings nevertheless have important applications to the efficient transmission of videos over bandwidth constrained or otherwise unstable connections. In this work, we develop a deep learning architecture capable of restoring detail to compressed videos which leverages the underlying structure and motion information embedded in the video bitstream. We show that this improves restoration accuracy compared to prior compression correction methods and is competitive when compared with recent deep-learning-based video compression methods on rate-distortion while achieving higher throughput.
49.BREAK: Bronchi Reconstruction by gEodesic transformation And sKeleton embedding ⬇️
Airway segmentation is critical for virtual bronchoscopy and computer-aided pulmonary disease analysis. In recent years, convolutional neural networks (CNNs) have been widely used to delineate the bronchial tree. However, the segmentation results of the CNN-based methods usually include many discontinuous branches, which need manual repair in clinical use. A major reason for the breakages is that the appearance of the airway wall can be affected by the lung disease as well as the adjacency of the vessels, while the network tends to overfit to these special patterns in the training set. To learn robust features for these areas, we design a multi-branch framework that adopts the geodesic distance transform to capture the intensity changes between airway lumen and wall. Another reason for the breakages is the intra-class imbalance. Since the volume of the peripheral bronchi may be much smaller than the large branches in an input patch, the common segmentation loss is not sensitive to the breakages among the distal branches. Therefore, in this paper, a breakage-sensitive regularization term is designed and can be easily combined with other loss functions. Extensive experiments are conducted on publicly available datasets. Compared with state-of-the-art methods, our framework can detect more branches while maintaining competitive segmentation performance.