After i install face_recognition, it shows illegal command

[wilicyy]

after i install face_recognition, it shows illegal command, pls teach me how to get it work?

thanks,

[AastaLLL]

Hi,

Not sure if you a similar issue as below:

Best way to face this problem

Hi all, I’m a CUDA newbie programmer. I’m writing here to ask you guys to share opinions and ideas. First of all the goal: to reduce a 100 x 256 x 4 floats vector to just a 4 floats one A cpu-computed equivalent might be expressed like follows: memset(output, 0, 4*sizeof(float)); for( int i = 0; i < 100 * 256 * 4; i++ ) { output[i%4] += input[i]; } Do you guys think that the reduce6 version of the SDK examples might be deployed in this case? How to choose threads’ and blocks’ number…

Would you mind to setup an environment variable and run the face_recognition again?

$ export OPENBLAS_CORETYPE=ARMV8

Thanks.

[wilicyy]

yes boss, it works after above command.
many thanks!
and if work RTSP by face_recognition, how can i output Chinese string name on video? it seems OPENCV doesn’t work chinese name.

[AastaLLL]

Hi,

You can use ImageFont from PIL to achieve this.
Please check below page for the detailed information:

How to draw Chinese text on the image using `cv2.putText`correctly? (Python+OpenCV)

Thanks.

[wilicyy]

thank you boss and i will try PIL.
another question is when is pull RTSP stream by 1920*1080 for face recognition, shows resize( ) error and stops. problem is here:
small_frame = cv2.resize(frame, (0, 0), fx=0.25, fy=0.25)
how can i get it work normally by RTSP?

thanks!

[AastaLLL]

Hi,

It’s recommended to use Deepstream SDK for multimedia usage.
You can find the RTSP support below:

https://docs.nvidia.com/metropolis/deepstream/dev-guide/text/DS_Overview.html#deepstream-graph-architecture

Thanks.