In collaboration with NEC, Skylark Labs deploys AI that automatically detects abandoned objects and suspicious loitering behavior in public spaces, providing security teams with early warnings before situations escalate.
Airports, train stations, and public plazas are inherently open environments where security teams must balance vigilance with the flow of thousands of daily visitors. Identifying genuinely suspicious behavior amid normal activity is one of the hardest problems in physical security. Bags and packages left unattended in crowded spaces are difficult to spot among the constant movement of people.
Distinguishing between someone waiting for a friend and someone conducting pre-operational surveillance requires context that traditional motion-based systems cannot provide. Major transit hubs see hundreds of thousands of people daily, making manual monitoring impractical. Existing systems generate excessive alerts, causing guards to ignore or dismiss warnings over time -- a phenomenon known as false alarm fatigue that degrades security posture across the entire operation.
"The hardest part of public space security is not detection itself -- it is detecting the right things. Our AI learns what normal looks like so it can reliably flag what is not."
Skylark Labs' AI builds a dynamic model of normal activity patterns for each monitored space. It then identifies deviations -- objects that appear and remain stationary, individuals who linger in unusual patterns -- and alerts security personnel with contextual information through the Kepler platform.
Abandoned object tracking identifies items that become stationary in the scene and determines whether their owner remains nearby or has departed. Loitering pattern analysis detects individuals who remain in a specific area beyond normal dwell times, accounting for context like waiting areas. The system continuously learns what is normal for each location, reducing false positives over time. All inference runs on edge AI hardware for real-time response.
Threat scores are assigned based on multiple behavioral indicators, ensuring security teams respond to the most critical events first. Every detection is logged with video evidence and timestamps through Skylark's Living Intelligence dashboard, supporting post-incident investigation and compliance auditing.
Context-aware detection eliminates the vast majority of nuisance alerts that plague traditional motion-based systems. Automated detection identifies suspicious activity in its early stages, giving security teams time to respond before escalation. A single system monitors dozens of camera feeds simultaneously, extending security coverage without adding staff.
The shift from reactive incident response to preventive threat identification transforms how public safety organizations protect open spaces. Every detection event creates an audit trail with timestamped video evidence, supporting both immediate response and long-term security analytics that continuously improve detection accuracy.
The NEC partnership addresses one of the most persistent challenges in public safety: identifying genuine threats in environments designed for open access. Skylark Labs' approach combines behavioral analysis with environmental context, delivering a detection capability that improves with deployment time rather than degrading with operator fatigue.
Discover how AI threat detection can protect your public spaces.
Schedule a Demo
