City-Scale AI with Digital Twins

Discover how our city-scale smart AI project utilizes NVIDIA Metropolis, video language models (VLMs), and AI agents to deploy a comprehensive set of real-time anomaly detection models in urban environments. From traffic flow monitoring and emergency vehicle recognition to roadwork and infrastructure safety, this session showcases how public-private partnerships drive successful implementations. By integrating DeepStream, TensorRT, Triton, NVIDIA AI Blueprint for video search and summarization, and MTMC (Multiple-Camera Tracking), alongside digital twin simulations, we ensure real-time data collection, continuous model retraining, and robust anomaly detection across dozens of urban scenarios.