search

ethnhe / PVN3D

Code for "PVN3D: A Deep Point-wise 3D Keypoints Hough Voting Network for 6DoF Pose Estimation", CVPR 2020
MIT License
482 stars 105 forks source link

Visualize on custom pictures/live video stream of the camera #102

Open RoyAmoyal opened 2 years ago

RoyAmoyal commented 2 years ago

Hey, I want to use your model on real live streaming with the YCB dataset. I have already managed to run it live on my Intel RealSense Camera.

Just to be sure it will work well on my camera:

  1. I have to modify the K parameter of camera instincts.
  2. The factor Depth is 1 for our camera because the real sense already returns the real distances from the camera but for some reason, it is working only with 1100, can't really understand why.

what else do I have to modify for our camera?

In Addition, where can I find the classes numbers of the YCB objects?

andreazuna89 commented 2 years ago

Hi @RoyAmoyal can you give some suggestions on how to customize the code for running live with a camera? Which part should I modify in the demo code? Because the data input for ycb dataset for example contains a lot of keys which I am not sure how to generate if I use my RGBD camera.

Thanks a lot

lucasbrodo commented 2 years ago

Hi @RoyAmoyal and @andreazuna89, I am facing the same issues. @RoyAmoyal Do you have code snippets or working code that you could eventually share with us? Thanks a lot