Adding YOLOX native support

[klemen1999]

Purpose

• Add pinned upstream YOLOX submodule and exporter for official .pth state-dict checkpoints.
• Support YOLOX Nano, Tiny, S, M, L, and X detection variants.
• Detect YOLOX using checkpoint state-dict markers rather than file extension.
• Export as yolov6r1, reusing its output names and grid decoder metadata.
• Configure YOLOX archive preprocessing as unscaled BGR input.
• Add focused YOLOX unit tests plus official checkpoint E2E coverage and rebalance full E2E shards.

Specification

None / not applicable

Dependencies & Potential Impact

None / not applicable

Deployment Plan

None / not applicable

Testing & Validation

None / not applicable

AI Usage

Assisted-by: AGENT_NAME:MODEL_VERSION [TOOL1] [TOOL2]

Submitted code was reviewed by a human: YES/NO

The author is taking the responsibility for the contribution: YES/NO

Summary by CodeRabbit

• New Features
  • Added support for exporting YOLOX detection models, including Nano, Tiny, S, M, L, and X variants.
  • YOLOX checkpoints are automatically detected and configured with appropriate architecture and preprocessing settings.
  • Class names are populated automatically for compatible COCO-trained models.
• Documentation
  • Updated the supported-models list and credits to include YOLOX.

[github-actions]

☂️ Python Coverage

current status: ✅

Overall Coverage

Lines	Covered	Coverage	Threshold	Status
2028	1130	56%	0%	🟢

New Files

No new covered files...

Modified Files

No covered modified files...

updated for commit: 9e800f1 by action🐍

[coderabbitai]

Warning

Review limit reached

@klemen1999, you've reached your PR review limit, so we couldn't start this review.

Next review available in: 9 minutes

Enable usage-based reviews in Billing to review now. Otherwise, wait until the next included review is available.
You're only billed for reviews past your plan's rate limits ($0.25/file).

How can I continue?

After more reviews become available, a review can be triggered using the @coderabbitai review command as a PR comment. Alternatively, push new commits to this PR.

To avoid repeated limits, reduce automatic review volume by pausing incremental auto-reviews earlier, using label-based review opt-in, excluding WIP or generated PR titles, or requesting reviews manually when the PR is ready. If your team needs uninterrupted high-volume reviews, an organization admin can enable usage-based reviews.

How do review limits work?

CodeRabbit enforces per-developer PR review limits for each organization. Most developers receive the normal plan review availability.

For paid Pro and Pro+ PR reviews, CodeRabbit uses adaptive limits for sustained high-volume activity. When a developer's recent PR review activity reaches the 95th percentile or higher among CodeRabbit users, additional reviews become available more gradually as earlier reviews age out of the rolling window.

Please refer docs for additional details.

Review details

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro Plus

Run ID: 351640b0-36b8-436d-a65f-3a2fef9e1a0b

📥 Commits

Reviewing files that changed from the base of the PR and between 2e72763 and 9e800f1.

📒 Files selected for processing (4)

• tests/e2e_shards.py
• tests/helper_functions.py
• tests/test_end2end.py
• tests/test_unittests.py

📝 Walkthrough

Walkthrough

YOLOX support was added across checkpoint detection, model construction, NNArchive export, CLI conversion, registry wiring, preprocessing, model catalogs, documentation, and end-to-end tests.

Changes

YOLOX conversion pipeline

Layer / File(s)	Summary
YOLOX checkpoint detection and architecture .gitmodules, pyproject.toml, tools/yolox/YOLOX, tools/version_detection/..., tools/yolox/yolox_exporter.py	YOLOX checkpoint markers and architecture inference were added alongside the YOLOX submodule and exporter dependencies.
YOLOX model and archive exporter tools/yolox/yolox_exporter.py, tools/modules/exporter.py	YOLOX detection outputs, checkpoint loading, model construction, class naming, parser configuration, and configurable preprocessing were added.
Registry and CLI integration tools/conversion_registry.py, tools/main.py, tools/version_detection/__init__.py	YOLOX exporter creation was registered, and omitted CLI encoding now defaults to BGR for YOLOX and RGB otherwise.
YOLOX test coverage and model catalog tests/..., README.md	YOLOX variants, checkpoint URLs, output checks, exporter tests, shard assignments, CLI choices, documentation, and updated shard expectations were added.

Estimated code review effort: 4 (Complex) | ~45 minutes

Sequence Diagram(s)

sequenceDiagram
  participant CLI as convert CLI
  participant Detection as detect_version
  participant Registry as conversion_registry
  participant Exporter as YoloXExporter
  participant Archive as make_nn_archive
  CLI->>Detection: detect YOLOX checkpoint
  Detection-->>CLI: return YOLOX_CONVERSION
  CLI->>Registry: create exporter for yolox
  Registry-->>CLI: return YoloXExporter
  CLI->>Exporter: export_nn_archive with BGR encoding
  Exporter->>Archive: generate NNArchive with YOLOExtendedParser

Loading

Possibly related PRs

• luxonis/tools#223: Both changes modify the CLI conversion flow and version-dependent telemetry behavior.
• luxonis/tools#224: Both changes modify Exporter.make_nn_archive.
• luxonis/tools#225: Both changes modify end-to-end shard assignments.

Suggested reviewers: rolandocortez

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 35.29% which is insufficient. The required threshold is 80.00%.	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title clearly summarizes the main change: native YOLOX support was added.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch feat/yolox

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[coderabbitai]

Actionable comments posted: 1

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In `@tools/yolox/YOLOX`:
- Line 1: Update the tools/yolox/YOLOX gitlink to a valid, fetchable commit SHA
from its configured YOLOX remote, preserving the intended release or branch.
Verify the pinned commit can be fetched and initialize the submodule
successfully.

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro Plus

Run ID: c03c4f0e-6310-465e-8b54-af5faa4e771d

📥 Commits

Reviewing files that changed from the base of the PR and between 0338cfc and 2e72763.

⛔ Files ignored due to path filters (1)

• media/coverage_badge.svg is excluded by !**/*.svg

📒 Files selected for processing (17)

• .gitmodules
• README.md
• pyproject.toml
• tests/conftest.py
• tests/constants.py
• tests/e2e_shards.py
• tests/nnarchive_output_checks.py
• tests/test_unittests.py
• tests/test_yolox_exporter.py
• tools/conversion_registry.py
• tools/main.py
• tools/modules/exporter.py
• tools/version_detection/__init__.py
• tools/version_detection/version_detection.py
• tools/yolox/YOLOX
• tools/yolox/__init__.py
• tools/yolox/yolox_exporter.py

[github-actions]

Test Results

  6 files    6 suites   14m 45s ⏱️
 32 tests  32 ✅ 0 💤 0 ❌
192 runs  192 ✅ 0 💤 0 ❌

Results for commit 9e800f1.

[rolandocortez]

LGTM