-
Notifications
You must be signed in to change notification settings - Fork 90
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
Added cell tracking functionality #140
Conversation
- Transformations applied to images in the image data generator are now synced across training examples. - Additional features are easier to include using the "feature" flag in the siamese_model & SiameseDataGenerator.
modified training, model_zoo, image_generators, and tracking to reflect the changes in the TrackingGeneratorTests notebook. changes were tested by training a new model. ** image_generators now imports keras_preprocessing ** In the newest version of tensorflow we can remove this import as tensorflow 1.11+ depends on keras_preprocessing. removed trailing whitespace tracking.cell_tracker now has a features argument to take in the same set of features as the siamese_model small updates to the tracking notebooks to make sure everything works properly.
confusion matrix is in TrackingTest and TrackingGeneratorTest notebooks training_siamese_model also takes in a seed parameter to be able to perform a confusion matrix on the validation set.
Now every part of the pipeline uses future frame
…epcell-tf into cell_tracking
…epcell-tf into cell_tracking
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.
-
The PEP8 tests are preventing the build from passing. I've made the required changes in PEP8 changes for cell_tracking #148 . Please check out the PR and merge it into cell_tracking at your convenience.
-
Ideally, I would also like to see a correctness test for the data generators (or at least any test where we can call flow() and have it run. Its commented out now because I do not know how to make the flow() work).
-
Finally, if it is not too much trouble, visualization in the notebook to show that the data looks good at the end of training would be helpful (like comparing raw data with tracked like on the README). It does not need to be a gif, but maybe compare 2 frames and show that they look like the same IDs before/after?
EDIT: the images from the notebook do not load well due to a video. We can clean this up in the future, but for now, I think the README gif is enough.
The correctness tests are still a problem but an issue will be added to track this.
* PEP8 changes. double quote converted to single quote. some variable name changes for line length * small update to tests but SiameseGenerator is still largely untested. * comment out siamese generator tests. needs proper fake-data.
Additional model added to identify and track cells from frame to frame while accounting for lineages through divisions. New class added to utilize the new data generators and models. New file type created to account for raw and tracked movies as well as lineage information.