Is it possible to run 2 videos simultaneously on ultralytics yolov3, Pytorch

Question

I'm wondering if it's possible to run 2 videos simultaneously as inputs of ultralytics yolo v3..

my goal is fuse the bbox of two videos... (data fusion)

but I'm curious if it is initially possible.

I'd like to ask anyone has tried something like this before..

trying to manipulate detect.py

Answer 1

Yes, it's possible to process two videos simultaneously using Ultralytics YOLOv3 with PyTorch.

Example code with python:

...
import multiprocessing
...
if __name__ == "__main__":
    video1 = 'path_to_video1.mp4'
    video2 = 'path_to_video2.mp4'
    
    p1 = multiprocessing.Process(target=process_video, args=(video1,))
    p2 = multiprocessing.Process(target=process_video, args=(video2,))
    
    p1.start()
    p2.start()
    
    p1.join()
    p2.join()
...

LoadStreams class in detect.py:

if webcam:
    video_paths = ['path_to_video1.mp4', 'path_to_video2.mp4']
    dataset = LoadStreams(video_paths, img_size=imgsz)
    bs = len(dataset)
elif screenshot:
...