My current approach is following a queue based system, adding new camera stream to a list and reading multiple streams into a queue by assigning camID to the streams. Anybody can help me with other standard industrial approaches for object detection.
Asked
Active
Viewed 327 times
0
-
You can just do it by multithreading programming, one thread for one camera. What is the used of the threads, it comes to solve the problem like yours. There is no need to read the camera into a queue and then read it back from the queue which slows down the speed. – Bahramdun Adil Mar 21 '19 at 02:00
-
hi @BahramdunAdil, yes I just worked on sample multithreading program for running more than one stream. Yea its a better approach than using queue method. [https://stackoverflow.com/questions/29664399/capturing-video-from-two-cameras-in-opencv-at-once] – Balaji Mar 21 '19 at 02:19
-
So you go for multithreading approach, first, you need to design a camera object and then for every camera you just need to create an instance of the camera and do start it. So every camera will run in its own thread, easier and faster than what you share the cameras all in one thread, which makes the logic more complex. – Bahramdun Adil Mar 21 '19 at 02:24