A Project of ISC Lab, Umich. Convert yolov8 model into a quantized one that runs on RK3588 for pose detection. Able to do pose detection on a single image or a live-stream video.
- Convert to .rknn model python3 convert.py [onnx_model_path] [platform] [dtype(optional)] [output_rknn_path(optional)]
platform choose from [rk3562,rk3566,rk3568,rk3588]
dtype choose from [i8, fp]
The size of the input should be (224,224) under i8(quantized mode)
example: cd into ./src and run python convert.py ../model/yolov8s-pose.onnx rk3588 fp
Requirements and wheel for rknntoolkit1.6.0 are provided in the repo.
About 20fps using a 60hz live cam in live demo.
For NPU performance: please refer to this note: https://www.notion.so/NPU-ML-On-Orange-pi-3f66c06106e945549b0cfa740b25fe6a
This project is based on the RKNN Model Zoo
(https://github.com/airockchip/rknn_model_zoo)