Closed rahulaggarwal965 closed 5 months ago
GT poses are obtained in LiDAR frame, whose frequency is slower than the camera frames. As data is synchronized, you can interpolate between the two frames to find a pose for the given frame.
Feel free to collaborate your interpolation method to the repo! Otherwise, we hope to have a ground truth interpolation for cameras added into the dataset soon.
If I wanted the associated ground truth poses for the "/ovc/rgb/data" rgb images, which indices should I use to index into the "Cn_T_C0" (I assume these are the camera transforms).
Right now, I have an RGB image array that is 2550xHxWx3 and a pose array (Cn_T_C0) of 1020x4x4. Given a timestamp t associated with RGB image A, do I have to find the two closest timestamps in the pose_gt.h5 "ts" dataset and interpolate between the corresponding poses to find the correct pose?