Real-time and Stable Skeleton-based models and tools

ReadMe will update.

1. HRNet weights in BaiduPans

   links：https://pan.baidu.com/s/1kA5YnB6ufaDxGnYoOyvujg codes：0my1

   put it in simple-HRNet/weights for demo put it in preprocess_data/simple-HRNet/weights for generate your own skeleton-based dataset.

2. Yolo weights in BaiduPans

   links: https://pan.baidu.com/s/1wWg1Zl35TaYKYNdXDkeaYw

   codes: d9vk

   put it in simple-HRNet/models/detectors/yolo/weights put it in preprocess_data/simple-HRNet/models/detectors/yolo/weights for generate your own skeleton-based dataset.

1. python 1video_to_image.py to get frames from videos
2. split dun frames into 5 parts python split5.py 3. cd simple-HRNet,
   1. python process_by_hrnet1-5 to generate json files from frames using hrnet(parallel).
   2. python generate_label, python kinetic_gendata.py to generate kinetic format custom dataset

code here:https://github.com/kenziyuliu/MS-G3D