-
-
Notifications
You must be signed in to change notification settings - Fork 1.8k
Fix bug dataset and metadataset fixes DeepLabCut/DeepLabCut#2390 #2405
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
Fix bug dataset and metadataset fixes DeepLabCut/DeepLabCut#2390 #2405
Conversation
- Remove unsued dataset path. - Fix getting metadataset. Now obtained from config rather than auxiliary function. - Suppress unused return values from load_metadataset. - Move load_metadata to where it is explicitly needed.
- Get metadataset file path from config rather than auxiliary function. - Use return_train_network_path function to get train/test pose_cfg paths - Move load_metadata to after try-catch block.
- Get metadataset file path from config rather than auxiliary function. - Suppress redefined TrainFraction from load_metadata function. - Use return_train_network_path function to get train/test pose_cfg paths - Move load_metadata to after try-catch block. - Fix minor naming/formatting issues
|
Thanks! |
- Fix handling of trainingsetindex="all" - Replace Try/Except on test/pose_cfg with check if modelfolder exists so it covers both train and test pose_cfgs. - Add TODO related to FileNotFoundError - Minor Refactoring
- Fix handling of trainingsetindex="all" - Replace Try/Except on test/pose_cfg with check if modelfolder exists so it covers both train and test pose_cfgs. - Add TODO related to FileNotFoundError - Minor Refactoring
|
I have looked for other areas of the code that use However, I also wanted to note explicitly that my changes now mean that the |
jeylau
left a comment
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Thanks @Tetra-quark! I'd only ask you to load the train image indices outside the loop, and that looks great!
- Load config and metadata outside nested loop.
|
Great, thanks! I think this is good to merge, and if some code refactoring is needed in evaluate/visualizemaps, that can be done in a separate PR @MMathisLab @Tetra-quark. |
n-poulsen
left a comment
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Some minor stye comments, otherwise looks good to me
Pull request to fix issue #2390
TODOs:
trainingsetindex='all'. Fix needs to be applied toevaluate.pyandmultianimal_evaluate.py.get_data_and_metadata_filenamesin other areas of the project and evaluate whether to replace with these with simple dict access from pose_cfg. (For file creation the method should be used but I think afterwards ideally it should be accessed from pose_cfg).dlc_cfgtotest_pose_cfgto be more specific and mirror the naming oftrain_pose_cfgtrain/pose_cfg.yamlis not found similarly to how load test config is in a try/catch block.