Input RGB or RGBD?

[LiYan0306]

I remember this algorithm can calculate the 6D pose of the object by inputting the RGB image and the distance between the target and the camera without inputting the RGBD information. how is this idea implemented? And what are the implementation ideas? My idea is that I can use the distance to generate a depth map of the target and fit the target to a circle with the centre of the target to generate the point cloud data of the target, is this feasible? Perhaps we can discuss this scenario in more detail.

[Goddard-Wen]

@LiYan0306 Hello, how is your research on the idea that FoundationPose only uses RGB output going? Could you share your progress? Thank you very much