Computer Vision / Video Analytics

NVIDIA Brings Low-Code Development to Vision AI with DeepStream 6.0

Flowchart of DeepsSream

DeepStream SDK 6.0 is now available for download. A powerful AI streaming analytics toolkit, DeepStream helps developers build high-performance, low-latency, complex, video analytics applications, and services. 

Graph Composer

This release introduces Graph Composer, a new low-code programming tool that enhances the DeepStream user experience. DeepStream and Graph Composer abstract the low-level platform, framework, and library specifics from the user, enabling developers to build accelerated applications within days rather than weeks.

Using a simple and intuitive UI, users construct processing pipelines using drag-and-drop operations. Compatibility and design rule checks built into the tool ensure that the pipeline is correct by construction. 

The NVIDIA extension registry provides components for all DeepStream plug-ins. Graph Composer can also create extensions automatically based on existing GStreamer plug-ins, allowing users to leverage their existing plug-in libraries. Templates are available to help users develop new custom extensions.

Container Builder

Developers can use the new Container Builder to create deployable containers for their applications in a single step. Container Builder can target deployments from the edge to the cloud, helping developers handle cross-platform migration.

To reduce development efforts and increase throughput, developers can use highly accurate pretrained models from NVIDIA TAO Toolkit and deploy them with DeepStream. 

With multiple programming options and new plug-ins, DeepStream can help developers of all backgrounds harness the power of AI.

Release Highlights

  • Graph Composer lets you assemble complex pipelines using an intuitive and easy-to-use UI and quickly deploy them with Container Builder.
  • Action Recognition helps you create pipelines to easily identify action in your scene with the new pre-processing plugin. Specify regions of interest and the number of frames you want to process simultaneously.
  • Audio-video synchronization enhances applications such as broadcasting and web conferencing.
  • Automatic Speech Recognition (ASR) is a new plug-in that performs automatic speech recognition on input audio data and outputs transcribed text.
  • Source code for the Python Bindings is now available on
  • Tracker Updates, including: 
    • Performance improvement; 
    • Easy integration of custom trackers; and 
    • New DeepSORT tracker.
  • Full REDIS support, previously alpha in version 5.1.
  • Over 20 plug-ins and 30 reference applications to help developers get started.

Getting Started

Discuss (0)