Closed happyiminjay1 closed 1 year ago
Reference points can be any distinguishable points on the object for which you know the 3D coordinates. Typical examples are object corners. During manual detection, you need to select those points in the RGB image. The detector is then able to compute the 6DoF pose.
I tried this manual detector but still could not make it work. After selecting 4 points on the image, the program just stopped. Do you have any ideas why? The 3D coordinates were selected in Meshlab and I assumed they are correct. Maybe the PNP does not work well enough?
Update: Increase the number of points to 6 making it work much better. Thank you for an amazing work!
Is there any guideline to get reference points?
Thanks.