A simple GUI for clicking on keypoints for training pose estimation models. Currently only supports one set of points per frame. Please consider contributing for multiple instances per frame!

• pip install pose-annotator

Customize default_config.yaml then launch using pose_annotator usr_cfg=path/to/custom/config.yaml

• Ctrl+S save
• Right next frame
• Left previous frame
• Down next keypoint
• Space next keypoint
• Up previous keypoint
• Delete delete keypoint

• Adding or deleting images from a directory with existing annotations
  • annotations are matched to image files by rank-order of the filename
• ModuleNotFoundError: No module named 'skbuild'
  • please pip install --upgrade pip

• Make player resize image appropriately on startup
  • current behavior: changing frame resizes images to fit in view; does not expand size on image resize
• support removing keypoints
• support multiple animals / people per image
• add action queue + undo button