Study On Fire Safety Management Of Building Construction Based On Image Recognition Technology

Due to the complex flow of construction site personnel,many flammable materials,frequent hot work and other factors,construction fire safety management has always been an important part of the construction safety management process.At present,the construction fire safety management includes manual patrol and workers’ fire safety knowledge education,but the low management efficiency based on manual patrol and the difficulty of traditional smoke sensing and temperature sensing equipment can not be used make the number of construction fire accidents continue to increase in recent years.Under the background of intelligent and industrialized construction industry,construction safety is very important to protect people’s lives and property.It is of great practical significance to carry out the research on construction fire safety management based on image recognition technology.In this paper,the image recognition technology is introduced into the scene of fire safety management in the construction site,and the image recognition technology is used to replace the manual patrol,so as to find the fire in the initial stage in time,improve the efficiency of fire safety management in the construction site,and reduce the loss of personnel and property in the construction fire accident.The main contents of this study are as followsIn this paper,based on the energy accidental release theory,system safety theory and cybernetics three accident safety management theory deduction,and combined with the characteristics of building construction fire safety management scene and traditional building construction fire safety management content and process,the realization theoretical path of building construction fire safety management based on image recognition technology is established.Secondly,through the analysis of the historical cases of building construction fire and related literature,this study determines the fire image recognition model training scene classification for building construction scene,and based on the above research of fire image recognition model training scene classification,through the transfer learning method training,obtains a fire image recognition model for building construction site fire safety management scene Model.On the basis of this model,according to the relevant requirements of fire safety management in construction site,a fire recognition system based on image recognition technology is built.Finally,the effectiveness of the system is verified from three aspects of system identification accuracy,operation robustness and real-time performance.The analysis of the research results shows that the image recognition technology can be used in the construction fire safety management scene,and the average fire recognition accuracy rate is more than 90%,which is higher than that of the existing research in this field.At the same time,the fire image recognition model and system are established by using the scene pre training classification method according to the characteristics of the construction scene,which improves the recognition accuracy and overcomes the influence of the existing research based on the empirical feature recognition method,which is difficult to avoid the fire light similarity and the change of the lighting environment in the construction site environment.It effectively improves the management efficiency of building construction safety management personnel to discover and control the development of fire accidents at the initial stage of fire,and provides a new way to reduce the loss of building construction fire accidents.