I have a RK3566 board with RKNN v1.5.0 and I'm trying to run object detection module using a YOLOv5 model. I tried downloading the onnx model provided here (https://github.com/airockchip/rknn_model_zoo/tree/main/examples/yolov5), and then convert to a rknn model to be used. However, when I tried running using the model, the result is very weird:
I have a RK3566 board with RKNN v1.5.0 and I'm trying to run object detection module using a YOLOv5 model. I tried downloading the onnx model provided here (https://github.com/airockchip/rknn_model_zoo/tree/main/examples/yolov5), and then convert to a rknn model to be used. However, when I tried running using the model, the result is very weird:
Actual Result:
Expected Result:
I suspect there is some issues with the conversion, maybe the version is wrong. I need to run yolov5n on an OrangePi 3b, but so far I only gotten the yolov5s rknn model provided in (https://github.com/rockchip-linux/rknpu2/tree/v1.5.0/examples/rknn_yolov5_demo/model/RK3566_RK3568) to be working. Please help, thank you.