SARAHAI-INFERENCE automates the tedious and error-prone task of constant monitoring, allowing operators to focus on response and decision-making.
Real-Time Object Detection: Utilizes the YOLOv8n model to identify and classify critical objects in real-time video feeds, including person, car, gun, knife, drone, and fire.
Advanced Color Recognition: Employs the CIEDE2000 color-difference formula to identify the specific color of detected objects, providing crucial descriptive details for reports and identification (e.g., "person in a red shirt")7.
Persistent Object Tracking: Assigns unique IDs to moving objects, tracking their position and calculating their speed across the field of view8.
Pattern of Life Anomaly (PoLA) Detection: Establishes a baseline of normal activity and automatically flags abnormal events, such as unusual object types for a given time of day or objects moving in unexpected locations or speeds9.