fill in the missing values of the sparse depth maps of kitti dataset

[YifeiAI]

Hi,

Thanks for posting your code!

I have a question for the depth maps in kitti dataset. You mentioned that before training you use the NYu tool box to fill in the sparse depth maps of kitti dataset. I am wondering:

1. Are the sparse depth image from the official website under the section "depth completion/depth prediction" with 14G?

2. Can you tell more specifically how to use the tool box to fill in the sparse depth map? And how accurate is it to use the filled-in depth map as ground truth label.

Thank you very much!

[haofengac]

Hello,

1. Yes, that is correct.
2. You can go to the github link given in the readme file and follow the instructions there. There is no guarantee about how close this is compared with the ground truth, but this toolbox is a convention used by most works. The scheme is based on this work as given by the website of the NYUv2 dataset.

[SuVanHuynh]

Hi xanderchf,
I followed your guides to fill the missing values of the sparse depth maps of kitti dataset (I gonna use it as my ground truth in training). I also used the NYU tool box but the result was too bad (the below picture).
Did you modify any parameters in the tool box to get the better results? If you did, could you please tell me which parameters?

Thank you very much!

[answerwwfthu]

@SuVanHuynh Could you please give more details about how to generate the result using tool box?

[victoriamazo]

Hi! Could you please explain how to convert exactly sparse to dense depth map? I have not found NYU tool box in python.

[LuizGuzzo]

Hi! Could you please explain how to convert exactly sparse to dense depth map? I have not found NYU tool box in python.

It's been a while since your question, but ialhashim made a python code that converts KITTI's sparse to dense.
https://gist.github.com/ialhashim/be6235489a9c43c6d240e8331836586a#file-fill_depth_colorization-py