princeton-vl/DPVO

There is a problem running evalutae_kitti.py

Opened this issue · 3 comments

微信图片_20241009160212
The check found that there was a problem with the value of the self_positions_xyz:
self._positions_xyz is : [[-0. -0. -0.]]
How can I fix this bug, check my data read, ground-truth is fine.

It seems like only one keyframe was created, meaning the model failed to initialize. This might occur if somehow the data was being read incorrectly.

If you comment-out the model inference and run with --show_img, do the images look correct?

Thank you for your response. I solved the issue by using the version of the code you provided two months ago.

FYI some of the cuda kernels were updated recently, so you may need to run
pip uninstall dpvo && pip install . to get the latest