A YOLOv5-based real-time monitoring system for fire access in residential buildings is designed

aiming at the problem of fire access occupancy that affects fire evacuation.An algorithm is designed to identify the categories of items occupying fire access

including vehicles

cardboard

sundries.Meanwhile

in order to meet the real-time monitoring requirements

the YOLOv5 network model is deployed on a Web website

allowing users to view the fire access video

equipment status

alarm status

and historical data in real time on mobile phones or computers.This system achieves a classification precision of over 95% in identifying items occupying the fire access

and an alarm reminder is issued once the fire access is occupied.It effectively achieves real-time monitoring of three different fire access areas in a residential building.