Fig. 4From: Learning 3D joint constraints from vision-based motion capture datasetsGenerated synthetic dataset for right hip joint with valid (green) and invalid (red) samplesBack to article page