Hi, thank you for your great and inspiring work!
Is it possible to train only the keypoint detection (e.g. from cvat cuboid annotations on 2D image), without any additional 3D information or other data/information?
I think that corresponds to the "Base Module" of your paper?
Hi, thank you for your great and inspiring work!
Is it possible to train only the keypoint detection (e.g. from cvat cuboid annotations on 2D image), without any additional 3D information or other data/information?
I think that corresponds to the "Base Module" of your paper?