EdgeYOLO is a PyTorch-based anchor-free object detection model optimized for edge computing scenarios. It supports deployment on multiple platforms including Nvidia Jetson, Huawei Ascend, Rockchip RK3588, and various inference runtimes like TensorRT, ONNX, MNN, RKNN, and OpenVINO. The project provides training code and pre-trained models achieving 50.6% AP on COCO2017 at real-time speeds.