Thanks for the wonderful paper and open-source project.
I noticed that not all files in the NYU Depth v2 (from sync.zip) were put in the training file list "nyudepthv2_train_files_with_gt.txt"; What were the criteria you used to select the image/depth pair in the training list?
For example for basement_001a, in the directory there are 201 RGB/Depth pairs; the training list only contains 134 of them; There are pairs such as rgb_00181.jpg and sync_depth_00181.png were not included;
What was the reason to choose only part of the image/ground truth pairs?
Hello;
Thanks for the wonderful paper and open-source project.
I noticed that not all files in the NYU Depth v2 (from sync.zip) were put in the training file list "nyudepthv2_train_files_with_gt.txt"; What were the criteria you used to select the image/depth pair in the training list?
For example for basement_001a, in the directory there are 201 RGB/Depth pairs; the training list only contains 134 of them; There are pairs such as rgb_00181.jpg and sync_depth_00181.png were not included;
What was the reason to choose only part of the image/ground truth pairs?
Thanks