I am currently working on 3D hand pose estimation. The idea is to first detect the 2.5D pose representation and then obtain 3D pose with the help of camera parameters. For some reason, I was trying to augment my dataset. But the augmentation libraries allow augmentation of 2D keypoints only. But my label is 2.5D i.e, x, y pixel co-ordinates and root relative normalised depth, z. After applying 2D keypoint augmentation, I became doubtful about whether this 2D only augmentation is not label preserving. I was wondering whether such an augmentation is valid and if it is whats a good way I can try out? TIA


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.