Skip to content

Latest commit

 

History

History

dnlnet

Folders and files

NameName
Last commit message
Last commit date

parent directory

..
 
 

Introduction

Official Repo

Code Snippet

DNLNet (ECCV'2020)
@misc{yin2020disentangled,
    title={Disentangled Non-Local Neural Networks},
    author={Minghao Yin and Zhuliang Yao and Yue Cao and Xiu Li and Zheng Zhang and Stephen Lin and Han Hu},
    year={2020},
    booktitle={ECCV}
}

Results

PASCAL VOC

Backbone Pretrain Crop Size Schedule Train/Eval Set mIoU Download
R-50-D8 ImageNet-1k-224x224 512x512 LR/POLICY/BS/EPOCH: 0.01/poly/16/60 trainaug/val 76.73% cfg | model | log
R-50-D16 ImageNet-1k-224x224 512x512 LR/POLICY/BS/EPOCH: 0.01/poly/16/60 trainaug/val 76.36% cfg | model | log
R-101-D8 ImageNet-1k-224x224 512x512 LR/POLICY/BS/EPOCH: 0.01/poly/16/60 trainaug/val 78.37% cfg | model | log
R-101-D16 ImageNet-1k-224x224 512x512 LR/POLICY/BS/EPOCH: 0.01/poly/16/60 trainaug/val 77.25% cfg | model | log

ADE20k

Backbone Pretrain Crop Size Schedule Train/Eval Set mIoU Download
R-50-D8 ImageNet-1k-224x224 512x512 LR/POLICY/BS/EPOCH: 0.01/poly/16/130 train/val 43.50% cfg | model | log
R-50-D16 ImageNet-1k-224x224 512x512 LR/POLICY/BS/EPOCH: 0.01/poly/16/130 train/val 40.68% cfg | model | log
R-101-D8 ImageNet-1k-224x224 512x512 LR/POLICY/BS/EPOCH: 0.01/poly/16/130 train/val 44.88% cfg | model | log
R-101-D16 ImageNet-1k-224x224 512x512 LR/POLICY/BS/EPOCH: 0.01/poly/16/130 train/val 41.80% cfg | model | log

CityScapes

Backbone Pretrain Crop Size Schedule Train/Eval Set mIoU Download
R-50-D8 ImageNet-1k-224x224 512x1024 LR/POLICY/BS/EPOCH: 0.01/poly/8/220 train/val 79.75% cfg | model | log
R-50-D16 ImageNet-1k-224x224 512x1024 LR/POLICY/BS/EPOCH: 0.01/poly/8/220 train/val 77.80% cfg | model | log
R-101-D8 ImageNet-1k-224x224 512x1024 LR/POLICY/BS/EPOCH: 0.01/poly/8/220 train/val 80.64% cfg | model | log
R-101-D16 ImageNet-1k-224x224 512x1024 LR/POLICY/BS/EPOCH: 0.01/poly/8/220 train/val 78.67% cfg | model | log

More

You can also download the model weights from following sources: