[Relax][ONNX] Add image.resize3d op and wire 5D Resize

[Dayuxiaoshui]

Summary

• Add Relax image.resize3d end-to-end: attrs, C++ op registration/inference, Python API, and legalization to topi.image.resize3d.
• Update ONNX 5D Resize to emit relax.image.resize3d instead of direct emit_te(topi.image.resize3d).
• Reuse the existing resize2d implementation pattern, which let us move faster while keeping behavior consistent and risk low.
• Add tests for op inference, TVMScript parser, legalization, ONNX import, and resize3d negative/error cases.

Test Plan

• python3 -m pytest -q tests/python/relax/test_op_image.py -k 'resize3d'
• python3 -m pytest -q tests/python/relax/test_transform_legalize_ops_image.py -k 'resize3d'
• python3 -m pytest -q tests/python/relax/test_tvmscript_parser_op_image.py -k 'resize3d'
• python3 -m pytest -q tests/python/relax/test_frontend_onnx.py -k 'resize_nd_sizes or resize_5d_emits_relax_resize3d'

[gemini-code-assist]

Summary of Changes

Hello, 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 enhances the image processing capabilities within Relax by introducing a new resize3d operator. This addition enables 3D image scaling, crucial for applications involving volumetric data. The integration with the ONNX frontend ensures that 5D Resize operations from ONNX models are correctly mapped to this new Relax operator, providing a more robust and native handling of such transformations. The implementation follows established patterns and is thoroughly tested to ensure correctness and stability.

Highlights

• New Operator: Introduced the relax.image.resize3d operator with full end-to-end support, including attributes, C++ registration, Python API, and legalization.
• ONNX Integration: Updated the ONNX 5D Resize operator to directly emit relax.image.resize3d, streamlining the import process.
• Consistent Implementation: Leveraged the existing resize2d implementation pattern to ensure consistency and reduce development risk.
• Comprehensive Testing: Added extensive tests covering operator inference, TVMScript parsing, legalization, ONNX import, and various error conditions for resize3d.

🧠 New Feature in Public Preview: You can now enable Memory to help Gemini Code Assist learn from your team's feedback. This makes future code reviews more consistent and personalized to your project's style. Click here to enable Memory in your admin console.

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.

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 introduces a new resize3d operator to the TVM Relax framework, extending image processing capabilities to 3D tensors. The changes involve defining new attributes for the resize3d operation, implementing its Python frontend and C++ backend, including operator registration, type inference, and layout inference. The ONNX frontend is updated to utilize this new operator for 5D tensor resizing, and comprehensive test cases have been added to validate its functionality and integration. A critical issue was identified in the resize3d C++ implementation where the argument count validation is incorrect, potentially leading to a crash due to out-of-bounds access if only one argument is passed, which also suggests a similar bug in the resize2d implementation.

[Dayuxiaoshui]

@tlopex Because there was a reference, the task was completed quickly. Next, I hope to take on the PyTorch fixes, but before that, I need to wait for this PR to be reviewed.

[tlopex]

Overall looks good to me, could you correct the point that gemini suggested?

[Dayuxiaoshui]

@tlopex We have resolved the issue and are awaiting approval.

[tlopex]

Now it looks good. Thank you @Dayuxiaoshui