Federal Project Grant Summary

The National Science Foundation's Division of Information and Intelligent Systems (CFDA 47.070) awarded $269,000 to the University of Illinois Urbana-Champaign on July 15, 2026, for a collaborative research project concluding June 30, 2029. This Computer and Information Science and Engineering (CISE) project, conducted in partnership with George Mason University, develops an artificial intelligence (AI)-driven framework for cross-camera video analytics that enables networked cameras to work collaboratively rather than as isolated systems. The research addresses real-world applications across infrastructure monitoring, building security, retail analytics, and laboratory environments where multiple cameras must coordinate to provide comprehensive scene understanding despite limited bandwidth and computing resources.

The project delivers three primary research contributions: (1) an asymmetric cross-camera video codec enabling separate camera encoding with joint server-side decoding and adaptive bandwidth management; (2) a unified inference framework with networked submodels that share features, intelligently schedule cameras, and support diverse analytics tasks; and (3) a reactive system adaptation mechanism utilizing adaptive buffer management to optimize video frame retention at the server. Rather than relying on rigid, handcrafted rules, the framework leverages machine learning to discover and integrate cross-camera dependencies into the end-to-end pipeline, achieving high accuracy and low latency under resource constraints typical of distributed surveillance and monitoring deployments.