Support YOLOX-v0.3.0 & ONNX Runtime support
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!
全てのバージョンにおいて、挙動はyolox_ros.pyを標準としています。すべてのソースコード(スクリプト)のメンテナンスは行っていないため、気になるところがあればissueなどで教えてください。
---更新---
yolo_ros_pyのデモプログラムをyolox_sからyolox_nanoに変更- ダウンロードされる重みの変更。以下は自動でダウンロードされる重み
- yolox_nano.pth
- yolox_nano.onnx
- ONNX Runtimeのサポート
yolox_ros_cppにおいてパラメータimage_size/widthとimage_size/heightの削除- この変更以降、
trtexecによる量子化が推奨され、torch2trtの使用は非推奨となりました。
- この変更以降、
- yoloxのpipインストール対応
English
I'm glad to get so many stars and forks after creating it. Thank you for your support.
If you can help me with GitHub Sponsors, it will encourage me to develop and maintain it!
In all versions, the standard behavior is yolox_ros.py The behavior is standard in all versions. I do not maintain all the source code (scripts), so if you have any concerns, please let me know via issues.
---Update---
- Changed
yolo_ros_pydemo program fromyolox_stoyolox_nano. - Change of downloaded weights. The following are the weights that are downloaded automatically
- yolox_nano.pth
- yolox_nano.onnx
- Support for ONNX Runtime
- Removal of parameters
image_size/widthandimage_size/heightinyolox_ros_cpp.- After this change, quantization with
trtexecis recommended and use oftorch2trtis deprecated.
- After this change, quantization with
- Support for pip installation of yolox
Supported YOLOX version
- YOLOX v0.3.0 ( release v0.3.0 )