Hi, I have a question about the ablation study:
You have evaluated the performance of HybridPose under different predicted intermediate representations on the LineMod dataset. Now would you like to provide more further information on code implementation. How can I only use keypoints or keypoints and symmetry in initialization and refinement sub-modules?
Thanks very much for your interest in our work! Please take a look at the exp_ablation branch of our repository. Around this line we have the code for initialization and this line for refinement.
I hope this helps! Let me know if you have further concerns.
Hi, I have a question about the ablation study: You have evaluated the performance of HybridPose under different predicted intermediate representations on the LineMod dataset. Now would you like to provide more further information on code implementation. How can I only use keypoints or keypoints and symmetry in initialization and refinement sub-modules?