Smoke Recognition and Localization in Building Corridors Based on Video Images
In order to realize the real-time detection of smoke spreading position in the building walkway,the smoke was identified by the target detection algorithm,and the two-dimensional position of smoke in the image was analyzed.Combined with the position re-lationship between smoke and the building walkway,a smoke front position location method based on monitoring video images was pro-posed.In the aspect of smoke front recognition,YOLOv7(You Only Look Once version 7)model was used as the smoke recognition model,and the model was trained by increasing data sets through data enhancement strategy,so as to get better recognition weight.In the aspect of smoke front position detection,the front smoke measurement method based on position information was adopted,and the two-dimensional information of smoke in the image was matched with the edge line of the roof of the walkway in the building.According to the mapping relationship between the image position information of the intersection of the smoke edge position and the edge line of the roof of the walkway in the building and the actual distance from the camera,the actual position of smoke in the image was obtained.The experimental test was carried out through the built walkway model,and the results show that the average error of the position of smoke front determined by this method is within 5％.The research results can provide support for using the existing building monitoring system to quickly grasp the smoke spread information,and provide basis for on-site fire rescue and personnel evacuation plan formula-tion.
building fireelongated building passagewaysmoke detectionsmoke location
NETL
NSTL
万方数据
