Vision Microservices

This page is for version 3.3.0 (default). For other versions, use one of these documentation indexes: - 3.3.0 (default): https://docs.nvidia.com/vss/3.3.0/llms.txt - Dev: https://docs.nvidia.com/vss/dev/llms.txt

The Vision Microservices extract rich visual features, semantic embeddings, and natural language understanding from video. Each is a standalone, GPU-accelerated microservice with documented REST and MCP interfaces, and can be deployed independently or composed into an agent workflow.

• Object Detection and Tracking - Real-time object detection, classification, and multi-object tracking on single or multi-camera streams.
• Real-Time VLM - Vision Language Models that generate captions, detect incidents, and identify anomalies in video streams.
• Video Embedding Generation - Semantic embeddings from video, images, and live RTSP streams for search and similarity matching.
• Video Summarization - Analysis and summarization of extended video recordings through chunking and aggregation of dense captions.