Does it support using real camera parameters and depth maps for 6D pose correction during inference?

[lkyw5210]

Thank you for your excellent work. I am working on 6D pose estimation based on the deploy.py file, and I have found that the rotation matrices predicted by the model contain errors. I would like to correct the rotation matrices using real camera intrinsics and depth maps, but I want to know if it is possible to perform this rotation matrix correction directly during inference without retraining the model with camera parameters. How should I modify the code?