[ViTDet] Remove hardcoded image preprocessing and add ViTDet ImageConverter

[DavidLandup0]

Description of the change

Currently, ViTDet has a hardcoded imagenet std/mean preprocessing step on the input image, which is not under a flag to turn on/off, and deviates from KHub's conversion convention.

This PR removes the hardcoded step and adds an ImageConverter for ViTDet.

Checklist

• I have added all the necessary unit tests for my change.
• I have verified that my change does not break existing code and works with all backends (TensorFlow, JAX, and PyTorch).
• My PR is based on the latest changes of the main branch (if unsure, rebase the code).
• I have followed the Keras Hub Model contribution guidelines in making these changes.
• I have followed the Keras Hub API design guidelines in making these changes.
• I have signed the Contributor License Agreement.

[gemini-code-assist]

Summary of Changes

Hello @DavidLandup0, 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 significantly refactors the image preprocessing for ViTDet models within Keras Hub. The primary goal is to remove the previously hardcoded ImageNet normalization directly embedded in the ViTDetBackbone, which deviated from Keras Hub's standard conversion conventions. By introducing a new, dedicated ViTDetImageConverter layer, the preprocessing logic becomes explicit, configurable, and aligns with the framework's modular design, offering better control over input transformations.

Highlights

• Preprocessing Refactor: The hardcoded ImageNet mean/std normalization step has been removed from the ViTDetBackbone class, addressing the issue of implicit preprocessing.
• New ImageConverter: A new ViTDetImageConverter layer has been introduced. This dedicated layer handles ImageNet normalization (mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225]) and scaling to [0, 1] for ViTDet models, making preprocessing explicit and configurable.
• API Exposure: The ViTDetImageConverter is now exposed through the keras_hub.layers API, allowing users to easily integrate it into their preprocessing pipelines.

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
Code Review	/gemini review	Performs a code review for the current pull request in its current state.
Pull Request Summary	/gemini summary	Provides a summary of the current pull request in its current state.
Comment	@gemini-code-assist	Responds in comments when explicitly tagged, both in pull request comments and review comments.
Help	/gemini help	Displays a list of available commands.

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 is a good architectural improvement, moving the hardcoded image preprocessing from ViTDetBackbone into a new, dedicated ViTDetImageConverter. The removal of the hardcoded logic in the backbone is done correctly. However, I've found a critical issue in the implementation of the new ViTDetImageConverter where it fails to apply the required ImageNet normalization due to incorrect arguments being passed to its base class. I have also included a suggestion to improve the clarity of the docstring example. Please see the detailed comments below.

[gemini-code-assist]

The example code is a bit misleading and incomplete:

1. It uses np without showing the import numpy as np statement.
2. np.random.rand() generates float values in [0, 1). The layer then scales this by 1/255, which is likely not the intended demonstration. Using np.random.randint(0, 256, ...) would better simulate a typical uint8 image for which scale=1.0 / 255.0 is appropriate.

Please add the import inside the ````pythonblock and userandint` for clarity.

Suggested change

	converter = keras_hub.layers.ViTDetImageConverter(image_size=(1024, 1024))
	converter(np.random.rand(1, 512, 512, 3)) # Resizes and normalizes
	converter = keras_hub.layers.ViTDetImageConverter(image_size=(1024, 1024))
	converter(np.random.randint(0, 256, size=(1, 512, 512, 3))) # Resizes and normalizes