search

Jiawei-Yao0812 / NDCScene

Official implementation for the ICCV 2023 paper "NDC-Scene: Boost Monocular 3D Semantic Scene Completion in Normalized Device Coordinates Space"
31 stars 3 forks source link

question about cam_pose and vox_origin #4

Open fdengmark opened 9 months ago

fdengmark commented 9 months ago

Hi:

I‘m trying to make some inferences using image capture from my own camera. However, I'm a little bit confuse about the cam_pose and vox_origin which are inputs when using NYUv2 dataset.

I try to find some documents discuss about it but I failed.

Here are my questions:

  1. where is the cam_pose in NYUv2 dataset comes from and what is it's world coordinate?
  2. what does vox_origin reqresents? How it is caluated?
  3. In nyu_dataset.py there is an parameter [self.img_D = [0.5538, 6.8243]]. What does it represents, and how does it being computed?

I'm not very familiar with NYUv2 dataset, so if you have some documents discuss about it would you share it? If not, can you explain it with short answer?

Thank you so much!

Jiawei-Yao0812 commented 5 months ago

Hi @fdengmark, sorry for the late reply as i have been busy recently. I hope your problem has been successfully solved.

If not, I strongly suggest that you could look through this website (https://cs.nyu.edu/~fergus/datasets/nyu_depth_v2.html) and which is very helpful!

If there is any questions or confusion, please feel free to let me know.