2026 EXHIBITOR DIGITAL SHOWROOM

This edge device receives video streams from existing network CCTV cameras and performs AI analysis on the incoming video. After detecting a person object, it conducts skeleton-based analysis to extract joint information and precisely analyzes posture and behavior to identify abnormal activities such as falling, trespassing, and violence. When smoke or flames are captured, it rapidly recognizes fire-related situations and performs an additional, more detailed analysis using a neural network model on the detected smoke or flame objects. Upon event detection, the identified regions are cropped and input into the VLM model within the AI Box, minimizing false alarms that may occur in surveillance operations.