Testing on other datasets

[salvadog]

Hi, I'm trying to test DPSNet on my own datasets. I've got several images together with their corresponding camera intrinsic and poses, but without depth maps. When I was testing them, error occurs that the .npy file is missing. So my questions are:
(1) Are .npy files the ground truth depth maps for the input images
(2) Are .npy files necessary for testing, or how can I run the testing process without .npy files.

Thanks a lot!

[sunghoonim]

Hi,
(1) Depth is ground truth depth map.
(2) You can run without .npy files. DPSNet does not require gt depth map. You need to modify the code not to load .npy file and not to compute th errors.

[salvadog]

Thanks for your timely reply!

[wykxyz]

hi I want to ask where the code of test_eth3d.Py?
the code for abs_rel','abs_diff','sq_rel','rms','log_rms','a1','a2','a3? @sunghoonim

[wykxyz]

If permiting. can we talk with wechat? my wechat wyk862180831 @sunghoonim

[muk3250]

@salvadog
Hello. Actually, I am also trying to test the performance of DPSNet on my own captured dataset as well. I used the COLMAP for reconstruction, and used camera intrinsic and poses from COLMAP as input to DPS but the output is not good. It seems to me that may be I need to rearrage/normalize the pose data extracted from COLMAP. May I ask how did you obtained/computed the camera poses and intrinstic for your dataset. Thanks in anticipation.