Can you tell me what methods I can use to simultaneously recognize 4 RTSP streams or more in real-time? Previously, I could do this with YOLOv8, but now with YOLOv9, I can't seem to do it; I start experiencing dropped frames around the second stream. I'm using a GPU 3060 with 12GB VRAM. Each of my RTSP streams has a frame rate of 30 frames per second. My operating system is Ubuntu 20.04, CUDA version 12.0, and NVIDIA-SMI version 525.60.13. I'm using the YOLOv9-c-converted.pt model. Below are images of 4 RTSP streams and a GPU snapshot. The 4 RTSP streams are from different cameras but within the same domain.
Can you tell me what methods I can use to simultaneously recognize 4 RTSP streams or more in real-time? Previously, I could do this with YOLOv8, but now with YOLOv9, I can't seem to do it; I start experiencing dropped frames around the second stream. I'm using a GPU 3060 with 12GB VRAM. Each of my RTSP streams has a frame rate of 30 frames per second. My operating system is Ubuntu 20.04, CUDA version 12.0, and NVIDIA-SMI version 525.60.13. I'm using the YOLOv9-c-converted.pt model. Below are images of 4 RTSP streams and a GPU snapshot. The 4 RTSP streams are from different cameras but within the same domain.