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 is a framework for building hardware-accelerated video and audio analytics pipelines. It simplifies the process of moving from a simple prototype (like a single-line command) to a production-grade edge AI deployment, reducing the amount of code required to implement complex media intelligence tasks.

How it works

Built on GStreamer and powered by OpenVINO, the framework allows users to define pipelines as simple strings or via Python/C++ code. These pipelines use specialized elements for inference, analytics, and output. It targets Intel CPU, GPU, and NPU hardware for optimized performance. It supports a wide variety of models from sources like Hugging Face, Ultralytics, and Geti, and can handle multi-stream, multi-sensor inputs including RTSP streams, LiDAR, and radar point clouds.

Who it’s for

It is designed for developers building edge AI applications, specifically those needing to integrate deep learning inference into media streams in real-time on Intel hardware.

Highlights

  • One-line pipelines: Create working detection pipelines using a single gst-launch-1.0 command.
  • Hardware acceleration: Unified codebase for Intel CPU, GPU, and NPU.
  • GenAI and VLM support: Ability to run Vision-Language Models (e.g., MiniCPM-V, CLIP, Whisper) within pipelines.
  • Extensibility: Custom logic can be added via Python callbacks or full Python GStreamer elements.
  • Integration: Built-in elements for publishing inference results directly to MQTT or Kafka.
  • Sensor Fusion: Supports multi-stream processing and 3D sensors like LiDAR and radar.

Related

  • Project
    insight-platform/Savant

    A high-level framework for building real-time, high-performance computer vision and video analytics pipelines on Nvidia hardware, abstracting the complexity of DeepStream.

  • Project
    nndeploy/nndeploy

    A high-performance AI deployment framework that uses visual workflows to deploy AI models across desktop, mobile, and edge devices.

  • Project
    NVIDIA-AI-Blueprints/video-search-and-summarization

    NVIDIA’s Video Search & Summarization (VSS) blueprint is a GPU‑accelerated reference stack that lets you search, query, and summarise live or recorded video with natural‑language. It combines vision‑language models, LLMs, and NVIDIA NIM micro‑services, offering real‑time analytics, alert verification, and long‑video summarisation via modular Python/Node/Next.js services and Docker‑Compose or cloud‑launchable deployment.

  • Project
    NVIDIA-NeMo/Curator

    A GPU-accelerated data curation framework for building repeatable pipelines to load, filter, deduplicate, and transform massive text, image, video, and audio datasets for AI training.

  • Project
    Visionary-Laboratory/holi-spatial

    A data curation pipeline that transforms video streams into annotated 3D spatial intelligence datasets, including 3D Gaussian Splatting geometry and spatial QA pairs.