Page 1 of 1

yolov5 rknn sample on pico ultra w

Posted: 2025-03-18 16:32
by steko
Hi,

I am evaluating yolov5 rknn sample. I've installed SDK and completed conversion of onyx->rknn. Development pc is connected to the board via add interface, and boards can reply pings.

I have transferred rknn server files to the board via add push:

Code: Select all

[root@luckfox root]# strings /oem/usr/bin/rknn_server | grep -i "rknn_server version"
rknn_server version: 2.3.0 (e80ac5c build@2024-11-07T12:53:13)
[root@luckfox root]# strings /oem/usr/lib/librknnmrt.so | grep -i "librknnmrt version"
librknnmrt version: 2.3.0 (c949ad889d@2024-11-07T11:37:43)
When I executed the model file after starting rknn server I received the following error:

Code: Select all

(venv) luckfox@luckfox:~/rknn_model_zoo/examples/yolov5/python$ python ./yolov5.py --model_path ../model/yolov5s_relu.rknn --target rv1106 --img_show
use anchors from '../model/anchors_yolov5.txt', which is [[[10.0, 13.0], [16.0, 30.0], [33.0, 23.0]], [[30.0, 61.0], [62.0, 45.0], [59.0, 119.0]], [[116.0, 90.0], [156.0, 198.0], [373.0, 326.0]]]
I rknn-toolkit2 version: 2.3.0
--> Init runtime environment
adb: unable to connect for root: closed
I target set by user is: rv1106
I Get hardware info: target_platform = rv1103, os = Linux, aarch = aarch32
I Check RV1106 board npu runtime version
E init_runtime: get target_platform is rv1106, while required target_platform = rv1103 
I also receive this debug message: "[root@luckfox root]# [ 220.147931] adbd (2117): /proc/2117/oom_adj is deprecate
d, please use /proc/2117/oom_score_adj instead."

How can I troubleshoot the adb connection error and the inaccurate target platform identification?



Kind regards,

Re: yolov5 rknn sample on pico ultra w

Posted: 2025-03-19 2:57
by Crocodile
Hello, we have not tested the rknnserver on the linked board, the platform name entered according to the prompt information and the detected platform name do not match, resulting in a running error, please try to use it

Code: Select all

 python ./yolov5.py --model_path .. /model/yolov5s_relu.rknn --target rv1103 --img_show