Deep High-Resolution Representation Learning for Human Pose Estimation (CVPR 2019)

My environment Ubuntu18.04,3090

when you download code from

https://github.com/leoxiaobin/deep-high-resolution-net.pytorch

Configure for the code

Upgrade your pip

python -m ensurepip

python -m pip install --upgrade pip

modify the requirement.txt

pip install -r requirements.txt

modified requirements.txt

EasyDict==1.7

pycocotools

opencv-python

shapely

#opencv-python==3.4.9.31

# shapely==1.6.4

Cython

scipy

pandas

pyyaml

json_tricks

scikit-image

yacs>=0.1.5

tensorboardX==1.6

Download the pre-trained model in the table,pretrained-model was split in three way by definition on datasets as follows.

https://onedrive.live.com/?authkey=%21AEwfaSueYurmSRA&id=56B9F9C97F261712%2111775&cid=56B9F9C97F261712

Get your file struture like this

and build a new folder "models/pytorch/pose_coco" to contain you pretrained models

 

Configure the pretrained_model path in Attribute PRETRAINED and  MODEL FILE

PRETRAINED: 'models/pytorch/pose_coco/pose_resnet_50_256x192.pth'

MODEL_FILE: 'models/pytorch/pose_coco/pose_resnet_50_256x192.pth'

My test image :1.jpg

 Because I use the pretrained model pose_resnet_50_256x192.pth,so I need to use file "experiments/coco/resnet/res50_256x192_d256x3_adam_lr1e-3.yaml" to configure and match my pretrained model.

python demo/demo.py --image 1.jpg --cfg experiments/coco/resnet/res50_256x192_d256x3_adam_lr1e-3.yaml  --showFps --write

Output.jpg

python demo/demo.py --video 2.gif --cfg experiments/coco/resnet/res50_256x192_d256x3_adam_lr1e-3.yaml  --showFps --write