Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Reproduce_Train on MegaDepth #27

Open
AubreyCH opened this issue Jun 2, 2023 · 0 comments
Open

Reproduce_Train on MegaDepth #27

AubreyCH opened this issue Jun 2, 2023 · 0 comments

Comments

@AubreyCH
Copy link

AubreyCH commented Jun 2, 2023

Hi!
Follow the LoFTR, I download the megadepth_v1.tar.gz from its official website. However, the [D2-Net preprocessed images] are not available anymore. As the author of LoFTR suggested, I can preprocess the megadepth following D2-Net or use megadepth_v1. Sadly, preprocessing megadepth as D2-net is memory-consuming, I CAN'T afford that. However, directly using megadepth_v1.tar.gz has a data loading error.

After printing the self.scene_info = np.load(npz_path, allow_pickle=True) from src/datasets/megadepth.py,
image

image

It shows that depth paths are in the same organization as in megadepth_v1, however, images should be in the Undistorted_SfM file, which I guess is the [D2-Net preprocessed images].

MAYBE, one solution is that

  1. create a dir called Undistorted_SfM
  2. put images from megadepth_v1 dataset to this dir like 'Undistorted_SfM/5003/images/00109.jpg',

It seems would work, I don't sure. Can somebody plz give me some suggestions?
OR, if someone can share D2-Net preprocessed images with me, that would be SO grateful!

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

1 participant