HandLandmark Detection that can be performed only in onnxruntime. Pre-focusing by skeletal detection is not performed. This does not use MediaPipe. - View it on GitHub