suggestion: make camera_opencv.py compatible with rpiOs Bullseye

1634718277

Hi
I use a PiCar Adeept_AWR.
I created a Sd card with RPI_os bullseye
The camera is declared non-legacy and therefore the camera_opencv.py program is incompatible with this configuration and must be modified using the new picamera2 library.
1 - install picamera2
https://github.com/raspberrypi/picamera2
2 - check libcamera installation
in a terminal enter "libcamera-hello"
3 - edit camera_opencv.py
add in header (I have it, put under import cv2)
from picamera2 import Picamera2
4 - At the end of the file modify the Camera class method
     @staticmethod
     def frames():

1. 
   
2.     @staticmethod
   
3.     def frames():
   
4. #        camera = cv2.VideoCapture(Camera.video_source) #,cv2.CAP_V4L2)
   
5.         camera = Picamera2(Camera.video_source)
   
6.         camera.configure(camera.create_preview_configuration(main={"format": 'XRGB8888', "size": (640, 480)}))
   
7.         camera.start()
   
8. #        if not camera.isOpened():
   
9. #            raise RuntimeError('Could not start camera.')
   
10. 
    
11.         cvt = CVThread()
    
12.         cvt.start()
    
13. 
    
14.         while True:
    
15.             # read current frame
    
16. #            _, img = camera.read()
    
17.             img = camera.capture_array()
    
18. 
    
19.             if Camera.modeSelect == 'none':
    
20.                 switch.switch(1,0)
    
21.                 cvt.pause()
    
22.             else:
    
23.                 if cvt.CVThreading:
    
24.                     pass
    
25.                 else:
    
26.                     cvt.mode(Camera.modeSelect, img)
    
27.                     cvt.resume()
    
28.                 try:
    
29.                     img = cvt.elementDraw(img)
    
30.                 except:
    
31.                     pass
    
32.             
    
33. 
    
34. 
    
35.             # encode as a jpeg image and return it
    
36.             yield cv2.imencode('.jpg', img)[1].tobytes()
    

Copy the Code

1750665706

Have you tried integrating more advanced image processing functions in cvt.elementDraw(img), like object detection or face recognition? I'm curious if Picamera2 has any performance impact when running in parallel with complex image processing threads?