Hi all,
I want to process multi-stream RTSP on jetson nano. I used H264 hardware decoder of jetson for this task with gstreamer + opencv in python. I don’t have problem in this step.
For capturing the frames of each streams, I used one thread per stream for capruring the frames. then I feeded batch frame of streams to model for processing. This way maybe cause a missing frames duo to processing time of model. one solution maybe to use thread
queue for this problem. I want to know, which of solutions can be effecisnt in jetson nano, thread of queue or redis ro kafka?
How solved the deep stream this problem?
Hi,
You may try to run default config file for Jetson Nano:
source8_1080p_dec_infer-resnet_tracker_tiled_display_fp16_nano.txt
And refer to the example to modify to RTSP sources:
Documents: