Migrate From DeepStream to DLStreamer

NVIDIA DeepStream and Intel DLStreamer are both end-to-end streaming analytics toolkits for building AI-powered video analytics and computer vision applications. They both enable developers to efficiently process and analyze multiple video, image, and sensor streams in real time.

If you have an existing solution relying on DeepStream and would like to see how you can benefit from DL Streamer, there are two options available:

• Introduce Intel DL Streamer to your pipeline to coexist with Nvidia DeepStream, without the need to commit to just one toolkit. See the coexistence guide.

• Migrate to Intel DLStreamer by adjusting your video analytics pipeline to use OpenVINO-based inference and DL Streamer’s GStreamer elements instead of NVIDIA’s TensorRT-based ones. See the migration guide.

DeepStream Element	DL Streamer Equivalent
nvinfer	gvadetect, gvaclassify, gvainference
nvstreammux	Removed / handled internally
nvtracker	gvatrack
nvdsosd	gvametaconvert + overlay logic

Here is an example of what the conversion may look like:

DeepStream Element

`nvstreammux` ! `nvinfer config-file-path=config.txt` ! `nvdsosd`

DL Streamer Equivalent

`gvadetect model=model.xml model-proc=model-proc.json` ! `gvametaconvert`