update u-net notebooks to Keras 3

[takumiohym]

Update summary:

• update u-net notebooks to Keras 3
• Removed flip data augmentation since it was applied to image and mask separately that break the feature-label mapping, and this augmentation itself is not relevant to the learning goal.

[review-notebook-app]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB

[gemini-code-assist]

Summary of Changes

Hello @takumiohym, I'm Gemini Code Assist¹! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed!

This pull request modernizes the U-Net segmentation notebook by migrating its codebase to Keras 3, ensuring compatibility with the latest deep learning framework. A key improvement involves the removal of a flawed image flipping data augmentation strategy that could lead to incorrect feature-label associations, thereby enhancing the reliability of the training process. Additionally, various environment and dependency versions have been updated to align with current best practices.

Highlights

• Keras 3 Migration: The U-Net segmentation notebook has been updated to use Keras 3 APIs, replacing tf.keras calls with keras and utilizing keras.ops for backend operations across various layers, models, and utility functions.
• Data Augmentation Refinement: The image flipping data augmentation, which was incorrectly applied to image and mask separately, has been removed from the training pipeline to prevent feature-label mapping issues and improve data integrity.
• Dependency and Environment Updates: The notebook now specifies oxford_iiit_pet:3.2.0 for the dataset, updates the Docker image for training to tf-gpu.2-17.py310, and reflects newer workbench notebook and Python versions for improved compatibility and performance.

Using Gemini Code Assist

The full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips.

Invoking Gemini

You can request assistance from Gemini at any point by creating a comment using either /gemini <command> or @gemini-code-assist <command>. Below is a summary of the supported commands on the current page.

Feature	Command	Description

Customization

To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a .gemini/ folder in the base of the repository. Detailed instructions can be found here.

Limitations & Feedback

Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for Github and other Google products, sign up here.

You can also get AI-powered code generation, chat, as well as code reviews directly in the IDE at no cost with the Gemini Code Assist IDE Extension.

Footnotes

1. Review the Privacy Notices, Generative AI Prohibited Use Policy, Terms of Service, and learn how to configure Gemini Code Assist in GitHub here. Gemini can make mistakes, so double check it and use code with caution. ↩

[gemini-code-assist]

Code Review

This pull request updates the U-Net segmentation notebook to use Keras 3 APIs, which is a great step towards modernization. The changes mostly involve replacing tf.keras with keras, which has been done correctly in most places. The PR also removes the data augmentation step from the training script, as explained in the description.

My main concern is a critical issue I've found in the data preprocessing logic. The switch from tf.image.convert_image_dtype to keras.ops.cast has removed the pixel value normalization to the [0, 1] range. This will likely cause the model to behave incorrectly during training and inference. I've added two comments with suggestions to fix this, one for the notebook cell and one for the train.py script.