CivicGuard

The Behavioral Distress Recognition System is an AI-powered surveillance solution designed to transform traditional CCTV cameras from passive recording tools into active safety guardians. Instead of merely capturing footage for post-incident review, this system continuously analyzes live video feeds to recognize early signs of human distress and potentially dangerous situations before a crime occurs.
Using computer vision and behavior analysis, the system detects pre-crime body language patterns such as persistent following (stalking), crowding or surrounding behavior, and sudden panic-driven flight responses. These subtle cues are often missed by human operators, especially in dense or fast-moving public environments like railway stations, campuses, malls, or streets.
When a distress behavior is identified, the system immediately generates an alert on a security dashboard, highlighting the individuals involved with bounding boxes and providing a short video clip of the detected event. This allows monitoring personnel to quickly assess the situation and intervene in real time—before physical contact or escalation happens.
The solution is designed with ethical and practical considerations in mind. It focuses purely on behavioral patterns rather than identity recognition, ensuring privacy while still enhancing public safety. By acting as an intelligent assistant to human security teams, the system improves situational awareness, reduces response time, and helps prevent incidents rather than just documenting them.
In essence, this project reimagines urban surveillance as a proactive safety mechanism, capable of identifying distress in crowded spaces and enabling timely, informed intervention.