Fire Detection Method in Complex Environment Based on Improved YOLOv5
Aiming at the problems of video image distortion and low detection accuracy caused by uneven distribution of dust in complex environment,a fire detection method based on improved YOLOv5 in complex environment is proposed.Firstly,the improved dark channel prior dehazing algorithm is used to dehazing the collected fire image to improve the recognition accuracy of fire video image in complex environment.Secondly,the CA(Coordinate Attention)attention mechanism is introduced into the YOLOv5 network model framework to enhance the flame features,suppress other useless features,and improve the efficiency and accuracy of fire detection.Finally,in order to solve the problem that YOLOv5 has a poor detection effect on small targets,a small target detection layer is added to the feature fusion part of YOLOv5 to improve the detection ability of small targets.The experimental results show that the accuracy of the improved YOLOv5 network model reaches 80.5%,which is 4.2%higher than that of the original YOLOv5 network model.At the same time,the improved YOLOv5 network model has higher detection accuracy for small targets and effectively improves the accuracy of fire recognition in complex environments.
万方数据
维普
