Models for Toolkit v2.3.2 (v2)

Note: All models and weights are subject to their license. The origin of each model is:
yolo-nas: https://github.com/Deci-AI/super-gradients
yolov9: https://github.com/WongKinYiu/yolov9
yolox: https://github.com/airockchip/YOLOX (which is a fork of https://github.com/Megvii-BaseDetection/YOLOX)

Models for Toolkit v2.3.2 (v1)

Warning: Some of these models don't work. See release v2.3.2-2.

Note: All models and weights are subject to their license. The origin of each model is:
yolo-nas: https://github.com/Deci-AI/super-gradients
yolov9: https://github.com/WongKinYiu/yolov9
yolox: https://github.com/airockchip/YOLOX (which is a fork of https://github.com/Megvii-BaseDetection/YOLOX)

Models for Toolkit v2.3.0

Note: yolo-nas models use pre-trained weights from DeciAI and are subject to their license. They can't be used commercially.

Models for Toolkit v2.0.0

Note: yolo-nas models use pre-trained weights from DeciAI and are subject to their license. They can't be used commercially.

v1.6.0

These are the rknn models converted from the default yolov8 models (from this repo https://github.com/ultralytics/ultralytics).

1. Follow instructions from https://github.com/ultralytics/ultralytics to export the yolov8 models to onnx format (imags size 320x320, opset 12)
2. Use RKNN Toolkit 2 to convert to rknn format (no quantization, optimizations set to 3 (default value)).

yolov8 models for rk3588

Version 1.5.2

yolov8 models for rk3568

Version 1.5.2

yolov8 models for rk3566

Version 1.5.2

yolov8 models for rk3562

version 1.5.2