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

Datamodule: some pending bits #126

Open
2 of 4 tasks
sfmig opened this issue Mar 18, 2024 · 0 comments
Open
2 of 4 tasks

Datamodule: some pending bits #126

sfmig opened this issue Mar 18, 2024 · 0 comments

Comments

@sfmig
Copy link
Collaborator

sfmig commented Mar 18, 2024

  • list of files to exclude: I would like to make the list of files to exclude to use wildcards / regex, rather than explicitly passing the names of the files.

  • train/test split: a true(er) test for generalisation would be to have different videos in the test set, and left and right videos always together in the same set (either train or test). Can we implement this, aka the train set to only have some videos (both Left and Right versions)? Robustness tests: non-overlapping test/train datasets #178

  • change CLI input arguments for evaluate_model.py script. Right now, we are passing directly the list of full paths to the image directories and the list of full paths to annotation files, which is not the most convenient. I think we can factor out and reuse the bits from the train_model.py script.

  • tracking the split_seed in MLFlow - I think right now it is not being logged correctly (any help here is more than welcome).

@sfmig sfmig self-assigned this Mar 18, 2024
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