Right now. i am trying to stream raspberry pi camera from other jetson nano and on the other jetson nano to run darknet object detection. Is there a way to stream the video and run darknet on the other jetson nano for detection?
Thank you
On the other Jetson Nano, you can run DeepStream SDK to receive/decode RTSP stream and do inferencing. Please refer to the document(6.0.1): NVIDIA Metropolis Documentation
Thank you for your kind assistance.
I’m able to run the stream with the commands that you sent. It turns out, the stream has a delay for around 3 seconds. Is there any way to decrease the latency to perform a better inference on jetson nano?
I did try UDP and it works well with latency around 500 milliseconds. Now, i’m trying to stream my pi camera video feeds but it always to show error like this:
$ gst-launch-1.0 videotestsrc is-live=1 ! video/x-raw,width=1280,height=720 ! timeoverlay valignment=4 halignment=1 ! nvvidconv ! ‘nvarguscamerasrc ! video/x-raw(memory:NVMM),width=1280,height=720’ ! tee name=t ! nvv4l2h264enc insert-sps-pps=1 idrinterval=15 ! h264parse ! rtph264pay ! udpsink host=192.168.0.101 port=5000 sync=0 t. ! queue ! nvegltransform ! nveglglessink sync=0
It Works! thank you so much. Now i’m trying to run darknet object detection on the other jetson nano. Is it possible to run darknet via UDP Stream? I think streaming udp is already part of the functionality of darknet, as in like to run something as: ./darknet detector demo data/voc.data cfg/yolo-voc.cfg yolo-voc.weights udp://@:5000
Or is there an actual way to perform the detection?
Thanks