Implementing an effective Graph Convolutional Neural Network for Classification of Human Pose
Run pip install torch
2D Human Pose Estimation: New Benchmark and State of the Art Analysis
- Create a folder that contain all the images
- Use human_pose.py for YOLOV8 human pose estimation
- Reformat the data from 17 keypoint to 16 keypoint
- torch_heavy_pose_classifier.py for training data using normal fully connected layer
- torch_robust_pose_classifier.py for training using GCN pose classifier
- ray_torch_pose_classifier.py using ray_tune for tuning
This repo is based on Kinetic-GAN
Thanks to the original authors for their work!