Create RetinaNet object detection guide

[sachinprasadhs]

Summary:
This Pull Request introduces comprehensive documentation detailing how to use an object detection model from Keras Hub, using RetinaNet as an example.

This tutorial includes:

• Loading and preprocessing a training dataset
• Demonstrating inference using pretrained weights
• Fine-tuning a pretrained object detector
• Custom training of an object detector
• Summary

[divyashreepathihalli]

Thanks for the PR @sachinprasadhs - since this is a guide - we should have more detailed explanations that would help users understand our framework better. Lets add more of that.

Thanks!

[mattdangerw]

Thanks! All minor comments. Also heads up I added keras-team/keras-hub#2219 for coco id - name mappings. Whenever it is released we can use it.

[divyashreepathihalli]

/gemini review

[gemini-code-assist]

Code Review

This pull request introduces a guide for object detection with RetinaNet using Keras Hub. I have identified a critical issue in the data loading logic, along with suggestions to improve code quality and clarity.

[sachinprasadhs]

@gemini-code-assist review

[gemini-code-assist]

Code Review

This pull request adds a new tutorial for object detection using RetinaNet and KerasHub, covering data preparation, inference, and fine-tuning. The implementation includes notebook, script, and markdown versions. Reviewers identified several issues, most notably a bug in the metadata builder that caused data leakage by loading all annotations instead of split-specific ones. Other feedback included a pathing error in the VOC data loader, the use of global variables in preprocessing functions, and a recommendation to separate backup and checkpoint directories for better file management. All suggested improvements aim to enhance the correctness and modularity of the tutorial code.