PAVE and Urban Safety Edge Analytics

Real-time video analytics for pedestrian safety over edge and end devices.

Role: Co-author and applied AI contributor
Recognition: Best Paper Award, ACM/IEEE Symposium on Edge Computing 2025

A moment from the COSMOS testbed camera at Amsterdam Avenue and West 120th Street, with danger zones recomputed for this figure: each moving vehicle's footprint swept along its predicted path for the next 2 seconds, and a pedestrian inside one. Phones compare their own position against the zones locally, so no personal data leaves the device. Bottom: per-frame latency of the edge pipeline on an NVIDIA A100, from Table 3 of the paper.

PAVE-style urban safety analytics combine street camera perception, edge computing, and end device alerts to support real-time pedestrian safety while preserving privacy.

The SEC 2025 paper received a Best Paper Award and demonstrates a scalable architecture for processing live video feeds, detecting pedestrians and vehicles, predicting vehicle trajectories, and sending anonymized danger-zone information to end-user devices.

Project details

Keywords: Edge video analytics

Trajectory prediction

Privacy-preserving warning systems

Real-time deployment

Pedestrian safety

City-scale sensing

References

2025

Real-Time Video Analytics for Urban Safety: Deployment over Edge and End Devices

Mahshid Ghasemi, Yongjie Fu, Xinyu Ouyang, Peiran Wang, Mehmet Kerem Turkcan, Jhonatan Tavori, Sofia Kleisarchaki, Thomas Calmant, Levent Gurgen, Zoran Kostic, Xuan Di, Gil Zussman, and Javad Ghaderi

In Proceedings of the Tenth ACM/IEEE Symposium on Edge Computing, 2025

Best Paper Award

DOI HTML

PAVE is a privacy-preserving real-time video analytics system that uses edge and end-device processing to support pedestrian safety at urban intersections.