基于YOLOv7和GMM算法的火焰实时检测方法OA

Most production sites in the petrochemical industry have the characteristics of flammability,explosion,high temperature,and high pressure,and are the core areas of safety supervision.Aiming at the problem that the current safety supervision area mainly relies on surveillance video playback and manual inspection,which leads to poor real-time performance and low security,a real-time flame detection method based on YOLOv7 and GMM algorithm is proposed in this paper.First of all,through the real-time data collection and analysis on the production site,based on the advantages of YOLOv7 in the field of target detection,the suspected flame area is detected;secondly,combined with the unique dynamic characteristics of the flame,the GMM model is used to eliminate the suspected flame area in the detection results,such as light,sunlight and so on;finally,the morphological operation method is used to further improve the accuracy of real flame detection.The experimental results show that this method can filter out the interference of suspected flame,detect the real flame effectively,and still has good anti-jamming ability and recognition accuracy under complex background.