#257 · Primary category: Computer Vision

DAMO-YOLO

damo-yolo deep-learning imagenet nas object-detection onnx pytorch tensorrt yolo yolov5

DAMO-YOLO: a fast and accurate object detection method with some new techs, including NAS backbones, efficient RepGFPN, ZeroHead, AlignedOTA, and distillation enhancement.

Project last updated:05/25/24

GitHub Stars

3.2K

Forks

407

Contributors

11

License

Apache-2.0

Why we included this project

DAMO-YOLO comes from Alibaba's DAMO lab as a YOLO-family detector aimed at real-time object detection in images and video. It pairs neural architecture search backbones with a lightweight feature network and head, which is how it keeps accuracy up without pushing latency out of reach. The repo ships several model sizes plus a Nano series that runs in real time on modest Intel CPUs, so you can pick the speed and precision trade-off that matches your hardware. It also covers the full path from training on a custom dataset to deploying via ONNX or TensorRT, with pretrained weights, a finetuning tutorial, and a quantization guide to get you started. If you are comparing detection frameworks for production, this is a practical one to test against your own data.

Articles for this project

No articles for this project yet.

To suggest a topic or contribute an article, contact us.

Related projects in this category