Closed minhlenhat21 closed 1 month ago
Since TUM don't have velocity, we calculate velocity based on first-order differentiation.
In /utils/deferentiate_vel.py
I noticed in TUMVI file config, the name dataset that you used in training is so small compare to total TUMVI dataset. Does it affect the result ?
If I remember correctly, only parts of the TUMVI dataset contain the ground truth within Vicon. @minhlenhat21
I try to download the TUM VI dataset but i can't find the "mav0/mocap0/grad_velo.txt" in there (your code load TUM VI has read this file) so basically i can't train with this dataset. Could you help me with that