for example, 16 cameras, only one GPU on a server, maybe at most init 4 graph to decode,inference, than encode. So every 1 graph need to process 4 video streams. but I didn't find any config like camera_id or source_id yet in mediapipe.
for example, 16 cameras, only one GPU on a server, maybe at most init 4 graph to decode,inference, than encode. So every 1 graph need to process 4 video streams. but I didn't find any config like camera_id or source_id yet in mediapipe.