Segmentation Fault When Loading Custom RKNN

[mdurrani808]

When running ./build-linux_RK3588 on an Orange Pi 5 running Ubuntu 22.04, I get a segmentation fault on line 26 (the line that runs rknn_yolov5_demo. This is a model converted from onnx using rknn-toolkit-2 if that helps. Any tips to get this working? Regular model provided works fine w/o segmentation faults.
2023.zip

Seems like it has the segfault on line 52 of rknnPool.hpp

[leafqycc]

When running ./build-linux_RK3588 on an Orange Pi 5 running Ubuntu 22.04, I get a segmentation fault on line 26 (the line that runs rknn_yolov5_demo. This is a model converted from onnx using rknn-toolkit-2 if that helps. Any tips to get this working? Regular model provided works fine w/o segmentation faults. 2023.zip

Seems like it has the segfault on line 52 of rknnPool.hpp

I guess you didn't specify the platform during the conversion process，
#rknn.config(mean_values=[[0, 0, 0]], std_values=[[255, 255, 255]])
rknn.config(mean_values=[[0, 0, 0]], std_values=[[255, 255, 255]],target_platform='rk3588')

and you can refer to this repository for model training
https://github.com/airockchip/yolov5