search

nianticlabs / manydepth

[CVPR 2021] Self-supervised depth estimation from short sequences
Other
607 stars 84 forks source link

Custom data sets (high speed scenes) do not work well #29

Open Klutch0615 opened 2 years ago

Klutch0615 commented 2 years ago

First of all, thank you for your contribution to the depth estimate! When I reproduced your code, I used my own custom dataset for training the monocular model, which consists of 6k consecutive frame images. I have also changed the intrinsics matrix K in the data loader, which should be correct after verification. However, the training results are still unsatisfactory, and I cannot even generate the correct depth map of the road, and I cannot get the correct depth of the vehicles on the road. So I would like to ask you what could be the cause of this situation? My personal guess is that apart from the relatively small dataset, is it possible that in high-speed scenarios where the environment is relatively simple, monocular training does not produce large losses and therefore the network cannot be trained sufficiently? I would be grateful for a solution! Thanks!

JamieWatson683 commented 2 years ago

Hi - very sorry for the delay!

Did you manage to get this working by any chance? If not, then could you post some example images from the dataset and some depth predictions?

Thanks a lot!