Hi all. In my current setup I use an RTSP adress as 3 individual sources with the DeepStreamSDK. Now, I am wondering if it is possible to assign a different model to each RTSP stream to inference real time (so 1 unique source has 1 unique model). The models I have are created with Azure CustomVision. So far I have only been able to find a way to run a primary and secondary inference on all of the incoming RTSP streams.
Are you using deepstream-app?
That is correct
You can run with multiple pipelines. The command line is like:
deepstream-app -c config1.txt -c config2.txt -c config.txt
Each config file contains one stream and one model.