Skip to content

README.md

Latest commit

 

History

History
53 lines (43 loc) · 2.34 KB

README.md

File metadata and controls

53 lines (43 loc) · 2.34 KB

Pretrained backbones

Introduction

Download the pretrained backbones from here:

  1. resnet50_mpii_pose.pth
  2. resnet50_coco_pose.pth
  3. hrnet_imagenet.pth
  4. hrnet_mpii_pose.pth
  5. hrnet_coco_pose.pth
  6. twins_svt_imagenet.pth
  7. twins_svt_mpii_pose.pth
  8. twins_svt_coco_pose.pth

Download the above resources and arrange them in the following file structure:

mmhuman3d
├── mmhuman3d
├── docs
├── tests
├── tools
├── configs
└── data
    └── checkpoints
        ├── resnet50_mpii_pose.pth
        ├── resnet50_coco_pose.pth
        ├── hrnet_imagenet.pth
        ├── hrnet_mpii_pose.pth
        ├── hrnet_coco_pose.pth
        ├── twins_svt_imagenet.pth
        ├── twins_svt_mpii_pose.pth
        └── twins_svt_coco_pose.pth

Results and Models

We evaluate trained models on 3DPW. Values are MPJPE/PA-MPJPE.

Backbones Weights Config 3DPW
ResNet-50 ImageNet resnet50_hmr_imagenet.py 64.55
ResNet-50 MPII resnet50_hmr_mpii.py 60.60
ResNet-50 COCO resnet50_hmr_coco.py 57.26
HRNet-W32 ImageNet hrnet_hmr_imagenet.py 64.27
HRNet-W32 MPII hrnet_hmr_mpii.py 55.93
HRNet-W32 COCO hrnet_hmr_coco.py 54.47
Twins-SVT ImageNet twins_svt_hmr_imagenet.py 60.11
Twins-SVT MPII twins_svt_hmr_mpii.py 56.80
Twins-SVT COCO twins_svt_hmr_coco.py 52.61