Link IP camera to object detection application

• Hardware Platform (Jetson / GPU) :NANO
• JetPack Version (valid for Jetson only): dsnano_v1-0-0_20GB_200131A
• TensorRT Version:6.0

I ran the 01_ObjDetect and it good. Now I want to detect cars in the street by an IP camera. However There is a problem about how to link plugins. I use playbin based on Gstreamer to get video stream from IP camera, here are the plugins in playbin:
It can connect to camera and show video to me.
I try to link uridecodebin with nvstrammux, (it is the forth plugins in 01_ObjDetect pipeline ) but error: Pipeline doesn’t want to pause

May I know how to use IP camera in deepstream?

The sample deepstream-test3 can be more close to the ucsecase. Please check


Thank u very much. I notice that there are many resnet10.caffemodel. Is it possible to change to caffe resnet18, or could I use my own pytorch resnet18 model?


You would need to convert the models to be TensorRT executable and then you can apply it.