Seamlessly Develop Vision AI Applications with NVIDIA DeepStream SDK 6.2

NVIDIA announced the general availability of the NVIDIA DeepStream SDK 6.2, an AI analytics toolkit for building high-performance video analytics and streaming applications. The update adds new features including enhanced multi-object trackers, support for new sensors, integration with REST APIs, updated Graph Composer, and enterprise-grade support through NVIDIA AI Enterprise.

Download the DeepStream SDK 6.2 now.

Release highlights

Access to multi-object trackers with state-of-the-art performance and accuracy

Multi-object trackers provide building blocks for vision AI applications where analyzing complex temporal changes of multiple moving objects is required. Version 6.2 offers three powerful tracker options under a unified architecture with significant feature improvements, for easy configuration and changes in different scenarios.

Integration with Basler cameras to enhance industrial inspection

Users now have easy integration between DeepStream and Basler machine vision GigE cameras with a new GStreamer source element plug-in and Pylon Viewer suite. With this approach, developers can use the Basler Pylon Viewer software to configure Basler cameras and generate configuration files for the DeepStream pipeline. A complete industrial inspection reference application with Basler camera support also makes it easy to understand, modify, and prototype.

Deploying SaaS Vision AI applications with DeepStream Rest APIs

With new REST-APIs for different plug-ins developers can create flexible applications that can be deployed as SaaS with controls from a simple web interface.

Additional enhancements:

  • Updated Graph Composer 2.5 with an easy-to-use user interface to assemble complex AI pipelines quickly.
  • Enterprise support available with NVIDIA AI Enterprise.
  • Support for LIDAR sensors and new LIDAR sample app.
  • New nvdsxfer plug-in enabling NVLink for data transfers across multiple GPUs.
  • New dewarper projections supported.
  • Ability to create custom bindings and scripts to install Python on any DeepStream containers.
  • Support for NVIDIA JetPack 5.1.
  • Support for NVIDIA Jetson Orin NX 16 GB.

Get started

Visit the DeepStream SDK 6.2 getting started page for quick start guides and instructional videos, including the following.

NVIDIA DeepStream technical deep dive showing DeepStream for inference combined with NVIDIA Triton and NVIDIA TensorRT

Review examples of the DeepStream SDK in action:

  • Fingermark, a company developing data-driven solutions for decision-making, uses DeepStream SDK in developing industrial vision AI solutions to improve worker safety.
  • Next-generation IT and business service provider, Trifork uses the DeepStream SDK to accelerate the development of the AI-based baggage tracking solution for airports.
  • Fyma, a software company providing computer vision AI platforms, uses DeepStream SDK to unlock the power of AI models, increase vision AI application performance, and reduce development time.

Debraj Sinha
Product Marketing Manager for Metropolis, NVIDIA

Here you’ll find a wealth of practical technical insights and expert advice to help you bring AI and visual intelligence into your products without flying blind.



1646 N. California Blvd.,
Suite 360
Walnut Creek, CA 94596 USA

Phone: +1 (925) 954-1411
Scroll to Top