DeepStream GS-Streamer

Please provide complete information as applicable to your setup.

• Hardware Platform (Jetson / GPU)
• DeepStream Version
• JetPack Version (valid for Jetson only)
• TensorRT Version
• NVIDIA GPU Driver Version (valid for GPU only)
Hello,
I need to build a pipeline with one source and two sinks , considering the first sink the frame will be fed to a custom algorithm, and the second sink will be Yolov3 Object Detector using DeepStream.

where in config file of yolov3 I can modify the source, since the available options are (uri , usb ,csi , … etc)

There is no algorithm in sink. Do you mean you need two nvinfer plugins(we defines it as GIE in our codes) in your pipeline? So what is the purpose of the two models(detection, classification , …)? Will the two models depends on each other? For example, the second model use the detected objects from the first model as the input.

If not, can you describe your purpose of your “a custom algorithm”?

Hello Fiona,
the first model is a normal object detection model, while the second algorithm is a lane detection algorithm using opencv classical techniques.

You may refer to deepstream-test2 sample codes for multiple nvinfer usage. If there is no dependency of the two models, the two models can both work in “Primary” mode.
https://docs.nvidia.com/metropolis/deepstream/dev-guide/index.html#page/DeepStream%20Plugins%20Development%20Guide/deepstream_plugin_details.html#wwpID0E0ZEB0HA