Study On The Application Of Pyrotechnic Detection Method Based On Deep Learning In Straw Banning

With the development of science and technology,the computing power of computers has gradually increased,and the scale of data has continued to expand.Artificial intelligence has entered a period of prosperous development.The application of artificial intelligence and deep learning has begun to penetrate into various industries and achieved good results.In recent years,the country has paid more and more attention to environmental protection.Relevant departments have pointed out that straw burning will cause serious environmental pollution,reduce air quality,and reduce soil nutrients,which have become environmental problems that need to be solved urgently.There are three main current straw burning monitoring methods: manual monitoring methods,sensor-based monitoring methods,and satellite remote sensing or drone-based monitoring methods.The manual monitoring method consumes a lot of human resources,has poor real-time performance and low efficiency.Sensor-based monitoring methods have technical limitations such as distance barriers,and are poor in timeliness,can not accurately provide early warning information.Based on satellite remote sensing or UAV monitoring methods,the collected images are susceptible to interference from environmental factors,and the processing of remote sensing images is not intelligent enough,nor can they play a role in early warning of straw burning.In recent years,artificial intelligence technology has made great progress in the field of image processing,and deep learning is one of the important branches.Using deep learning technology to solve the problem of firework detection and recognition has become another popular research direction.Therefore,in this paper,by studying artificial intelligence technology,the deep learning-based pyrotechnics detection technology is applied to the field of straw burning monitoring,and the straw burning monitoring system based on this method is constructed.The main contents of this paper are as follows:(1)This paper proposes a straw burning detection method based on smoke and fire detection,which is based on the YOLOv3 target detection algorithm,correspondingly improves and optimizes the straw burning phenomenon to achieve the goal of detecting the straw burning phenomenon.The specific implementation is as follows: Based on the YOLOv3 target detection algorithm,using the experimental platform and data set in this article,the straw detection model and the firework detection model are trained respectively,then the straw burning detection model is generated through the model fusion method,finally through the fused straw burning detection model to monitor the phenomenon of straw burning.In terms of experiments,the four common target detection algorithms(Faster R-CNN,R-FCN,SSD,YOLOv3)were tested separately,the YOLOv3 algorithm was determined as the basic algorithm through comparison,and its superiority was verified;in addition,Due to the small number of data sets,the effectiveness of data enhancement strategies(cutting,flipping,etc.)are verified through comparative experiments;finally,the three models in this article have tested their effects and verified their effectiveness.(2)A straw incineration monitoring system based on this method was constructed through the straw incineration detection method based on firework detection proposed in this paper.The entire straw burning monitoring system is mainly divided into four parts,namely the front-end layer,the transmission layer,the back-end layer and the monitoring platform layer.They are connected to each other to form a system to realize the purpose of straw burning monitoring.Then the functions and construction of the four modules are elaborated in detail.Finally,the platform of straw burning monitoring system is introduced...