Savant-AI Launches Open Source Computer Vision Framework for Real-Time Video Analytics on NVIDIA Hardware
Savant, an open-source Python framework for real-time computer vision, today announced the expansion of its developer ecosystem and released new case studies. Built specifically for NVIDIA edge devices and datacenter GPUs, from Jetson Orin Nano to H200, Savant makes it easier to create scalable video analytics systems using Python.
“Computer vision should not be a maze of integrations and GPU bottlenecks,” said Marika Mendelbaum, CEO of Savant. “We created Savant so developers can build deployable, scalable, and real-time systems using Python, without diving into low-level C++ or fighting with deployment hell.”
What is Savant
Savant is an open-source video analytics framework optimized for NVIDIA hardware. It provides:
-
The versatile, developer-friendly tools for the development of advanced computer vision systems, covering all aspects of working with video data.
-
Prebuilt components streamlining practical computer vision tasks like accessing various video data sources and storage, control flow management, video stream archiving, and visualization;
-
Unified codebase that runs on both Jetson edge devices and data center GPUs without changes;
-
Low-latency, high-throughput performance powered by DeepStream and TensorRT;
-
Batteries-enabled Python-first ecosystem familiar to data engineers.
Use Case: Factory Safety
A Savant-powered factory monitoring system was deployed to improve workplace safety. It:
-
Detected people and vehicles entering restricted zones
-
Verified workers were wearing protective equipment such as hard hats
-
Tracked movement to prevent accidents
-
Ran fully on local hardware with real-time alerts to the human supervisors
Other Applications
-
Smart city analysis systems, transforming visual information into measurable insights by person and vehicle traffic detection, classification, and recognition, including ANPR and complex event analysis.
-
Tracking solutions for sea, railroad, and automotive logistics companies.
-
Private video processing in privacy-sensitive environments, such as elder care.
Why Savant
-
Optimized for NVIDIA GPUs and Edge devices, and relies on the best Nvidia software technology, DeepStream and TensorRT.
-
Scalable for real production use, not just prototyping.
-
Handles live video streams from RTSP, USB CSI-2, and GigE Vision cameras, as well as video files.
-
Open source technology without commercial limitations.
-
Backed by an active community.
Get Involved
Developers are invited to explore the demos, contribute on GitHub, or join the community on Discord.
GitHub: https://github.com/insight-platform/Savant
Discord: https://discord.com/invite/KVAfGBsZGd
Media Contact
Company Name: Savant
Contact Person: Marika Mendelbaum
Email: Send Email
City: Boca Raton
State: FL
Country: United States
Website: https://savant-ai.io
