open-edge-platform/dlstreamer

Deep Learning Streamer (DL Streamer) Pipeline Framework is an open-source streaming media analytics framework, based on GStreamer* multimedia framework, for creating complex media analytics pipelines for the Cloud or at the Edge.

What it solves

DL Streamer simplifies the creation of hardware-accelerated video and audio analytics pipelines. It removes the complexity of writing extensive code for edge AI by allowing developers to build production-grade intelligence pipelines—ranging from simple object detection to multi-sensor deployments—using simple strings or minimal Python/C++ code.

How it works

Built on the GStreamer framework and powered by OpenVINO, DL Streamer provides a set of specialized elements (plugins) that can be chained together to form a processing pipeline. These pipelines can target Intel CPUs, GPUs, and NPUs from a single codebase. It supports a wide range of models from sources like Hugging Face, Ultralytics, and Geti, and can integrate directly with messaging systems like MQTT and Kafka for outputting inference results.

Who it’s for

It is designed for developers building edge AI applications, specifically those targeting Intel hardware, who need to deploy video analytics, 3D sensor processing (LiDAR/Radar), or Vision-Language Models (VLMs) with high performance and low development overhead.

Highlights

  • One-line pipelines: Create functional detection pipelines using a single gst-launch-1.0 command.
  • Broad hardware acceleration: Unified support for Intel CPU, GPU, and NPU.
  • GenAI and VLM ready: Capable of running Vision-Language Models like MiniCPM-V, CLIP, and Whisper.
  • Multi-sensor support: Ability to mux and demux multiple RTSP streams, LiDAR frames, and radar point clouds in one process.
  • Python-first extensibility: Allows adding custom logic via Python callbacks or full Python GStreamer elements.