About 6d representation

[gen4u]

Thanks for your great work.

I have a question about training the rotation matrix in the ScoreNet model.
According [1], for rotation matrix 6d representation is used, but when I look at the Genpose training code, it seems that the loss is applied in the 6d representation space, not in the original space. Is there a significant difference between applying loss in these two ways?

[1] Yi Zhou, Connelly Barnes, Jingwan Lu, Jimei Yang, and Hao Li. On the continuity of rotation representations in neural networks. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 5745–5753, 2019.