Lost in Time: Temporal Analytics for Long-Term Video Surveillance

TL;DR

This paper introduces methods for analyzing long-term video surveillance data to understand patterns and predict anomalies using heatmaps, footmaps, and trajectory-based models over a year of footage.

Contribution

It presents two novel schemes for descriptive and predictive temporal analytics tailored for long-term surveillance video data.

Findings

Heatmaps and footmaps reveal spatial-temporal patterns.

Trajectory and time-series models effectively predict anomalies.

One-year data demonstrates practical insights and anomaly detection capabilities.

Abstract

Video surveillance is a well researched area of study with substantial work done in the aspects of object detection, tracking and behavior analysis. With the abundance of video data captured over a long period of time, we can understand patterns in human behavior and scene dynamics through data-driven temporal analytics. In this work, we propose two schemes to perform descriptive and predictive analytics on long-term video surveillance data. We generate heatmap and footmap visualizations to describe spatially pooled trajectory patterns with respect to time and location. We also present two approaches for anomaly prediction at the day-level granularity: a trajectory-based statistical approach, and a time-series based approach. Experimentation with one year data from a single camera demonstrates the ability to uncover interesting insights about the scene and to predict anomalies reasonably well.