Hi:
I run pedestrian_tracker_demo error
has some error below
My version is latest version "l_openvino_toolkit_p_2019.1.144"
$ ./pedestrian_tracker_demo -i /home/asrock/inference_engine_samples_build/intel64/persion-detection-retail-0013/PexelVideos1338590.mp4 -m_det /home/asrock/inference_engine_samples_build/intel64/persion-detection-retail-0013/person-detection-retail-0013.xml -d_det CPU -m_reid /home/asrock/inference_engine_samples_build/intel64/person-reidentification-retail-0031/person-reidentification-retail-0031.xml -d_reid CPU
InferenceEngine:
API version ............ 1.6
Build .................. custom_releases/2019/R1.1_28dfbfdd28954c4dfd2f94403dd8dfc1f411038b
[ INFO ] Parsing input parameters
Loading plugin CPU
API version ............ 1.6
Build .................. 23780
Description ....... MKLDNNPlugin
Segmentation fault (core dumped)
How should I do??
Thank you!
連結已複製
Hi EdwinH,
Like the other posts, I'll move this one to computer vision forum from the OpenCL* forum to get more visibility from OpenVINO stakeholders.
You may want to post a stacktrace to the thread... The first thing I'd check for a seg fault for a program that many other users have executed successfully are the correctness of input arguments.
-MichaelC
Dear Hung, Edwin,
I tried it on both GPU and CPU (OpenVino 2019R1.1), and it worked just fine.
Here is my command:
pedestrian_tracker_demo.exe -i c:\users\sdramani\Downloads\sample-videos\people-detection.mp4 -m_det c:\users\sdramani\Downloads\person-detection-retail-0013.xml -m_reid c:\users\sdramani\Downloads\person-reidentification-retail-0031.xml -d_det CPU -d_reid CPU
The only difference between yours and mine is that you're using a different video stream. Can you try it with the people-detection.mp4 from https://github.com/intel-iot-devkit/sample-videos ? If you study the code (main.cpp) for the pedestrian_tracker_demo the video processing happens after plugin is loaded. I only get the following message once even though the plugin is called through -d_det and -d_reid twice :
Loading plugin CPU
API version ............ 1.6
Build .................. 23780
Description ....... MKLDNNPlugin
And the next part is the video processing. So my guess is that this is where you core dump.
Can you give one of the sample videos a try ? Please report back here. To figure out why your video doesn't work would require stepping through the code. I'm pretty sure it's not a model issue - but likely the sample was not built to work with your video.
Thanks,
Shubha
Dear Hung, Edwin,
IP Cameras are not optimal for the OpenVino samples. USB cameras work a lot better. It's possible but you may experience significant lagging and performance issues.
Please revisit IP Camera With OpenVino thread for the discussion.
Hope it helps,
Thanks,
Shubha
