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

About cityscapes datesets #23

Closed
a610402863 opened this issue Aug 21, 2017 · 4 comments
Closed

About cityscapes datesets #23

a610402863 opened this issue Aug 21, 2017 · 4 comments

Comments

@a610402863
Copy link

In the link about Cityscapes, I didn't find a data set called leftImg8bit_sequence_trainvaltest.zip. Is it a change of name or other reason?

@Yuliang-Zou
Copy link

Yuliang-Zou commented Aug 21, 2017

Seems that they changed the way they organize the dataset. I cannot use the preprocessing code for CityScapes either.

After modifying the code a little bit, I now have 1638 images as training split, 191 images as val split. Not sure if the author used the same setting.

@tinghuiz
Copy link
Owner

What kind of error does it give you when you use the preprocessing code?

@Yuliang-Zou
Copy link

What I did to make the code works:

  1. Remove the suffix "_sequence"
  2. Swap the orders of "snippet_id" and "tgt_local_frame_id" and orders of "snippet_id" and "curr_local_frame_id" in is_valid_example() and load_image_sequence() in data/cityscapes/cityscapes_loader.py

Then I can get 1638 images as training split, 191 images as val split.

@tinghuiz
Copy link
Owner

You need to download the package called leftImg8bit_sequence_trainvaltest.zip(324GB) so it contains video snippets. After preprocessing I had ~69K training snippets and ~7K validation snippets.

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

3 participants