Bae, Gwangbin

Hi, In `data/dataloader_scannet.py`, there is a code that replaces the image that does not exist. ``` if os.path.exists(scene_dir + '/color/{}.jpg'.format(img_idx + i)): img_idx_list.append(img_idx + i) else: img_idx_list.append(img_idx - i -...

I am sorry for the delayed response. I tested the same code, but it does not give me an error. Does the error happen for all the scenes? or does...

Hi, sorry for the delayed reply. Do you wish to test on multi-view images or single-view images?

Hi, sorry for the delayed response. Yes, this should be fixed. You can either remove that line or add a line `gt_dmap = F.interpolate(gt_dmap, size=[img.shape[2], img.shape[3]], mode='nearest')` FYI, they are...

Hi, thanks for your interest. Yes, this can happen. But in practice, the estimated 'mu' is positive (usually > 1m) and the scale of 'sigma' is significantly smaller than that...

Hi, D-Net is trained to estimate the depth map in absolute scale. If the scale of the test scene is significantly different from the scale of the training scenes, the...

Hi, very sorry for the delayed response. For NLL_ours, it is natural that the loss becomes negative. The likelihood can be higher than 1 and the loss (negative log likelihood)...

Hi, have you tried training on the full NYUv2 dataset which consists of 30K images?

@BayMaxBHL Hi, sorry for the delayed reply. I took GeoNet's data and saved them into uint8 png type by doing ((n+1) * 0.5) * 255.

@BayMaxBHL 1) For obvious reasons, test images were not included in the training set 2) Yes I also save two log files 3) When evaluating on NYUv2, which image/GT are...