Alexander

@ujgygy656565 Did you use a pretrained FlowNet?

try flownets_EPE1.951.pth.tar. Also it seems that your NN is undertrained.

It doesn't really matter which camera to use as an input.

@sunnyHelen try to rerun preprocess.py to update the files in pose_GT folder. The changes you mentioned affected the structure of the files.

@sunnyHelen It is normal.

@kourong try to clean up 'datainfo' folder and run main.py again.

@sunnyHelen It would be incorrect to directly compare mse_rotate and mse_translate with the metrics in the article because they are calculated differently. The authors of the article operated with RMSE...

@zhaihongjia you can download it directly from the site: http://www.cvlibs.net/datasets/kitti/eval_odometry.php

@yxh1993 try to clean up 'datainfo' folder and run preprocess.py and main.py again.

Hello @sunnyHelen , These numbers are the minimum and maximum sequence length. One of the approaches is to use random sequence lengths during training.