Thanks a lot! Adrian has great tutorials!
On my "Face Recognition Project", I really keep the 30 FPS (and not use threads) because I used for video capturing the standard videoCapture() (from cv2) to better explain the overall concept of Facial Recognition. But, for example on the project: "Object detection and tracking with OpenCV"', I have used the Adrian's Imutils library, where he has already optimized the FPS. But the final result should be only a little bit better because I am using imshow() to show the result, what decreases the fps optimation that was got using videoStream() (from Imutils) for video capturing instead of the standard videoCapture () (from cv2). And the 1ms delay that we should wait for program termination also complicates the overall optimization. As Adrian commented on the tutorial that you sent me, in the case a real project, we always can take out imshow() and waitKey(1), what should improve the performance and so reducing the latency a lot!
Thanks
Marcelo
Engineer, writer and forever student. Passionate to share knowledge of electronics with a focus on IoT and Robotics.