NVIDIA Released DeepStream 9.1: Bringing Agentic AI to Vision AI With 13 Skills and Multi-View 3D Tracking
NVIDIA released DeepStream 9.1, introducing 13 agentic skills for building video analytics pipelines. The update also features Multi-View 3D Tracking and AutoMagicCalib for simplified camera calibration.
- DeepStream 9.1 introduces 13 agentic skills for building video analytics pipelines
- Multi-View 3D Tracking enables accurate tracking across multiple cameras
- AutoMagicCalib automates camera calibration, simplifying the development process
NVIDIA's DeepStream 9.1 is a significant update to its video analytics platform, bringing agentic AI capabilities to vision AI. The 13 new skills enable coding agents like Claude Code and Codex to build complex video analytics pipelines from natural-language prompts.
The Multi-View 3D Tracking feature is a major highlight, allowing for the fusion of per-camera detections into a single, globally consistent 3D world. This is made possible by the assignment of a unique object ID, enabling more accurate tracking across multiple cameras.
The update also includes AutoMagicCalib, a feature that automates camera calibration, removing the need for manual intervention. Additionally, DeepStream 9.1 adds support for JetPack 7.2 and a unified open-source GitHub monorepo, making it easier for developers to access and contribute to the platform.
The release of DeepStream 9.1 demonstrates NVIDIA's commitment to advancing the field of computer vision and AI, providing developers with the tools they need to build innovative applications.
provides new tools and features for building innovative computer vision applications
enables the development of more accurate and efficient video analytics solutions
advances the field of computer vision and AI
- agentic AI
- a type of AI that enables coding agents to perform tasks autonomously
SQLite Query Explainer
AI ToolsDeterministic Data Engineering With AI Harnesses: Using Claude Code, Codex, Antigravity, and OpenCode for Data Work You Can Actually Trust
AI ToolsLoop Engineering with Adaptive PDF Parsing: Start Cheap, Pay for a Heavier Parser Only When the Page Needs It
AI ToolsFLUX on a 4070 Graphics Card 🖼️
Acura Rolls Out Google Gemini AI Assistant To Vehicles With Google Built-In - autoevolution
Oxmoor Valley residents protest Birmingham AI data center - wvtm
Residents of Oxmoor Valley are protesting the construction of a Birmingham AI data center. The protest is likely due to concerns over the data center's impact on the community.
AI ResearchDave Eggers told OpenAI staff that ChatGPT was ‘silencing an entire generation’
Author Dave Eggers told OpenAI staff that ChatGPT is silencing a generation, criticizing the AI model's impact on writers and artists. He expressed his concerns during a talk at the company's offices.
Meet Indian-origin tech leader Uday Ruddaraju, who has become OpenAI's CTO of Compute - The Times of India
OpenAI has named Indian‑origin tech executive Uday Ruddaraju as its new Chief Technology Officer of Compute.
LLMKimi: Threat or menace?
Moonshot AI, a Chinese startup, launched an updated version of its Kimi large language model this week. The release has sparked discussion about the potential for "AI communism" and broader geopolitical implications.
Bunkerhill raises $55M to scale agentic AI across health systems - AI News
Bunkerhill has raised $55M to expand its agentic AI solutions across health systems. This investment will enable the company to scale its operations and improve healthcare outcomes.
Chinese AI Model Takes U.S. by Surprise with Abilities Rivaling Claude and ChatGPT - Broadband Breakfast
A new Chinese artificial intelligence model has demonstrated performance capabilities comparable to top US systems like Claude and ChatGPT, surprising industry observers.