Process multiple video feeds ansyc about yolov7_openvino_cpp-python HOT 5 CLOSED

Hi you can create one async queue to load all of the feeds, or create multiple async queue to load each feed's request separately.
You can check this PR, see if it can help you on how to create multiple async queue.

from yolov7_openvino_cpp-python.

Gosh that is a bit over my head haha....is there anything funny with this approach below?

Am still tinkering around with the IoT rest API app to process 10 (or more) different IP cameras at once:

• One rest API endpoint for each camera
• Use multiprocessing to instantiate 10 instances of Python class you wrote YOLOV7_OPENVINO(object) or run them with the operating system as different .py files?
• std.out flush people detections if people count in frame was different than last frame and processed video frame something like this.

        sys.stdout.buffer.write(frame)
        sys.stdout.flush()

I idea comes from which is an outdated Intel course I took a while back with the link I showed you. Python code processes the video and std.out flushes everything to a separate web app running to build the dashboard.

The web app standpoint I have to keep track of the name of which video feed is getting processed to be able to build out the dashboard, am still wrapping my head around this...regardless if I can get some code to work if you have the time curious to see from a high level or not if things can be optimized better. Thanks!

from yolov7_openvino_cpp-python.

Not sure how this will work on live camera feeds but this is working so far with multiprocessing, this line where the ffmeg will flush in the draw method.

Its working on processing 3 video files at once : ) Am hoping to someday do something like this on GPU !

from yolov7_openvino_cpp-python.

Actually I had created a decode+ inference demo for multiple input data on GPU in C++. When you implement the codes in Python, we can leverage Async-queue to cache multiple inference request from different feeds. The logic of it looks like the diagram below:

If you can share your hardware architecture with me, I may provide you more suggestions on it.

from yolov7_openvino_cpp-python.

This is really cool will do...thanks @OpenVINO-dev-contest !

from yolov7_openvino_cpp-python.