CamSort AI

Security teams managing large facilities face a real problem. When you have hundreds or even thousands of cameras, it is impossible for a small team to monitor all of them at once. Most camera management software just shows you a wall of feeds and leaves the prioritization entirely up to the human operator. CamSort solves this by introducing an autonomous AI agent that continuously analyzes camera snapshots and ranks them based on what the operator actually cares about at that moment. The operator types a natural language instruction such as "sort by how hazardous the scene appears" or "find cameras with people present" and CamSort takes it from there. The system sends each camera snapshot independently to Google Gemini, which returns a score, a classification, a reason, and a recommended action for each feed. The backend then sorts all cameras by score and updates the dashboard automatically every few seconds without any manual input. CamSort also includes an operator priority layer that flags any camera crossing a critical threshold, making sure urgent situations are never buried under lower priority feeds. The live dashboard shows ranked camera cards with scores, reasons, and operator actions so the team always knows exactly where to look first. The project is built on Vultr, uses Google Gemini for multimodal image analysis, and is deployed with a live demo accessible to judges.