To train NYUv2, but the path name in txt is diffrent with dataset from BTS？

[jay25208]

Thank you for great work. How can I prepare NYUv2 for train and test. NYUv2 has 795 pics for training. Why your training pics number is 36253 in: --filenames_file data_splits/nyudepthv2_train_files_with_gt_dense.txt Or can you open your trianing data of NYUv2 as show in nyudepthv2_train_files_with_gt_dense.txt?

Why your test txt has 518.8579, is training and test use focal? --filenames_file_eval data_splits/nyudepthv2_test_files_with_gt.txt bathroom/rgb_00045.jpg bathroom/sync_depth_00045.png 518.8579

When I training the NYUv2 model，files are not found. python iebins/train.py configs/arguments_train_nyu.txt

In arguments_train_nyu.txt, the train and test are different, official_splits is from BTS. What is the meaning of sync. --data_path datasets/nyu_depth_v2/sync/ --gt_path datasets/nyu_depth_v2/sync/ --data_path_eval datasets/nyu_depth_v2/official_splits/test/ --gt_path_eval datasets/nyu_depth_v2/official_splits/test/

Why the path name in txt(nyudepthv2_train_files_with_gt_dense.txt) is diffrent with dataset from bts. Should I prepare my own txt(nyudepthv2_train_files_with_gt_dense.txt)?

For example, the path "bedroom_0130" is not in dataset from BTS. "bedroom_0130/rgb_00000.jpg bedroom_0130/sync_depth_00000.png bedroom_0130/dense/sync_depth_dense_00000.png" Why inputs need two depth map, or only use one? What's the difference between sync_depth_00000.png and sync_depth_dense_00000.png.

Thank you!

[jay25208]

Dataset is from bts , https://github.com/cleinc/bts/tree/master/pytorch