安装与运行命令不在主 README,需跳转 yolox_ros_cpp/README.md 查看;需按所选后端(TensorRT/OpenVINO/ONNXRuntime/TFLite)自行准备对应推理库;默认输出消息类型为 bboxes_ex_msgs,用 darknet_ros_msgs 需手动替换。
YOLOX-ROS
Ar-Ray-codeApache-2.0活跃维护把 YOLOX 目标检测封装成 ROS2 节点,支持 ONNX Runtime、OpenVINO、TensorRT、TFLite 等多种推理后端,方便机器人项目直接接入实时检测。
YOLOX + ROS2 object detection package (C++ only support)
安装与演示步骤见 yolox_ros_cpp/README.md,主 README 未给出具体命令README 未说明显存、CPU 或具体 GPU 型号要求;仅从后端列表可推断 TensorRT 与 ONNXRuntime CUDA 需 NVIDIA GPU,MIGraphX 需 AMD GPU,OpenVINO 与 TFLite 可走 CPU 路径。
把 YOLOX 目标检测接入 ROS2 的 C++ 包,订阅图像话题并发布检测框
把 YOLOX 目标检测封装成 ROS2 节点,支持 ONNX Runtime、OpenVINO、TensorRT、TFLite 等多种推理后端,方便机器人项目直接接入实时检测。
核心能力
- 订阅 sensor_msgs/Image 图像话题做推理
- 发布 bboxes_ex_msgs/BoundingBoxes 或 vision_msgs/Detection2DArray 检测框
- 支持 TensorRT、OpenVINO、ONNXRuntime CUDA/MIGraphX、TFLite 多种后端
- 可替换为 darknet_ros_msgs 消息类型以兼容现有流程
- 仅提供 C++ 实现,基于 ROS2 Jazzy
容易踩的坑
- v0.5.0最新正式版2026-01-03 · 2026-01-03 08:21 UTC
What's Changed
- add migraphx support by @Ar-Ray-code in https://github.com/Ar-Ray-code/YOLOX-ROS/pull/57
Full Changelog: https://github.com/Ar-Ray-code/YOLOX-ROS/compare/v0.4.1...v0.5.0
- v0.4.12024-09-23 · 2024-09-23 03:22 UTC
查看更新说明
What's Changed
- Preprocess speed up and some fix by @fateshelled in https://github.com/Ar-Ray-code/YOLOX-ROS/pull/54
- support ubuntu24 (but there is not tested) by @Ar-Ray-code in https://github.com/Ar-Ray-code/YOLOX-ROS/pull/55
Full Changelog: https://github.com/Ar-Ray-code/YOLOX-ROS/compare/v0.4.0...v0.4.1
- v0.4.02024-08-12 · 2024-08-12 04:43 UTC
查看更新说明
What's Changed
- Fix to suppress warnings of YOLOX by @m12watanabe1a in https://github.com/Ar-Ray-code/YOLOX-ROS/pull/48
- Replace TensorRT deperecated function by @fateshelled in https://github.com/Ar-Ray-code/YOLOX-ROS/pull/49
- rm yolox_ros_py by @Ar-Ray-code in https://github.com/Ar-Ray-code/YOLOX-ROS/pull/38
- add ci_rolling.yml (Ubuntu24) by @Ar-Ray-code in https://github.com/Ar-Ray-code/YOLOX-ROS/pull/51
- Feat/update params and interface by @Ar-Ray-code in https://github.com/Ar-Ray-code/YOLOX-ROS/pull/52
New Contributors
- @m12watanabe1a made their first contribution in https://github.com/Ar-Ray-code/YOLOX-ROS/pull/48
Full Changelog: https://github.com/Ar-Ray-code/YOLOX-ROS/compare/v0.3.3...v0.4.0
- v0.3.3v0.3.3 Support Humble2023-07-15 · 2023-07-15 09:42 UTC
查看更新说明
- Python is unsupported (Deletion is scheduled for the next ubuntu LTS)
- Update dockerfile for ROS 2 Humble
What's Changed
- Fix/dockerfiles by @Ar-Ray-code in https://github.com/Ar-Ray-code/YOLOX-ROS/pull/37
Full Changelog: https://github.com/Ar-Ray-code/YOLOX-ROS/compare/v0.3.2...v0.3.3
- v0.3.2v0.3.2 TFLite-C++ support2022-12-30 · 2022-12-30 04:13 UTC
查看更新说明
Japanese
作成後、多くのスターおよびフォークを頂けてうれしい限りです。ありがとうございます。
GitHub Sponsorsで支援して頂ければ開発とメンテナンスの励みになります!
English
We are very happy to receive many stars and forks since its creation. Thank you very much.
Please support us on GitHub Sponsors to encourage development and maintenance!
What's Changed
- update docs about YOLOX_ROS_CPP by @swiftfile in https://github.com/Ar-Ray-code/YOLOX-ROS/pull/23
- yolox_ros_cpp inference speed up. by @fateshelled in https://github.com/Ar-Ray-code/YOLOX-ROS/pull/24
- Support ONNXRuntime C++ by @fateshelled in https://github.com/Ar-Ray-code/YOLOX-ROS/pull/26
- support tflite C++ by @fateshelled in https://github.com/Ar-Ray-code/YOLOX-ROS/pull/31
- Update package.xml by @Ar-Ray-code in https://github.com/Ar-Ray-code/YOLOX-ROS/pull/33
New Contributors
- @swiftfile made their first contribution in https://github.com/Ar-Ray-code/YOLOX-ROS/pull/23
Full Changelog: https://github.com/Ar-Ray-code/YOLOX-ROS/compare/v0.3.1...v0.3.2
用户点评
登录用户的真实使用记录,可按版本和硬件筛选还没有人点评。第一个说出真实体验的人最有价值。
YOLOX-ROS
YOLOX + ROS2 Jazzy demo

Supported List
| Base | ROS2 C++ |
|---|---|
| TensorRT | ✅ |
| OpenVINO | ✅ |
| ONNXRuntime CUDA | ✅ |
| ONNXRuntime MIGraphX | ✅ |
| TFLite | ✅ |
Installation & Demo (C++)
Check this URL.
Topic
Subscribe
- image_raw (
sensor_msgs/Image)
Publish
- bounding_boxes (
bboxes_ex_msgs/BoundingBoxesorvision_msgs/Detection2DArray)bboxes_ex_msgs/BoundingBoxes: Output BoundingBoxes like darknet_ros_msgs- ※ If you want to use
darknet_ros_msgs, replacebboxes_ex_msgswithdarknet_ros_msgs.
Reference

@article{yolox2021,
title={YOLOX: Exceeding YOLO Series in 2021},
author={Ge, Zheng and Liu, Songtao and Wang, Feng and Li, Zeming and Sun, Jian},
journal={arXiv preprint arXiv:2107.08430},
year={2021}
}