Research And Implementation Of System On Intelligent Engineering Supervision Through Deep Learning For GTDecoration Project

Intelligent supervision is used in the finished residential interior decoration project of the built system.Through intelligent supervision,it can achieve the purpose of reducing labor costs,standardizing acceptance standards,and improving acceptance reliability.At present,the domestic decoration supervision project management relies on manual development,there are problems such as unreasonable human resource scheduling,lack of professionals,non-standard acceptance standards,and different acceptance results from person to person.Based on the study of deep learning,this paper intends to use computer vision technology,combined with microservices,to design and implement an intelligent supervision system,which is applied to the GTDecoration project;the system uses the intelligent supervision auxiliary module to identify supervision pictures and retrieve relevant acceptance standards to assist supervision Personnel to carry out work to improve the above-mentioned problems in the supervision of decoration projects.This article mainly does the following work:(1)According to analysis and discussions on the development of the monolithic architecture to the microservice architecture,the overall architecture of the supervision system and the intelligent detection auxiliary module can be designed by combining the front-end and back-end separation ideas.Meanwhile,complete the construction of the system framework.(2)Research the supervision target detection in the intelligent supervision auxiliary module.Based on the convolutional neural network,in view of the shortcomings of the SSD model,the MP-SSD algorithm is proposed to improve to obtain the target information in the supervision picture,and provide query conditions for the retrieval and acceptance criteria.(3)Optimize data storage and query,use Hbase to solve a large number of unstructured data storage problems in the supervision project;introduce a cache mechanism and use Elasticsearch to build a multi-level index to solve the problem of slow standard fuzzy query.(4)Realize the intelligent supervision system of decoration engineering.Through the demand analysis of the GTDecoration supervision project,the system function modules are clarified and the database design is carried out,and the realization of the mobile terminal and the web terminal module business is completed,and the system is verified to ensure the feasibility of the intelligent supervision system.The research results show that the use of a micro-service system architecture with separation of front and back ends can improve the scalability and maintainability of the system;through the improved algorithm MP-SSD based on SSD,the problem of missed inspection and re-inspection of supervision targets is effectively improved,and the recall rate is increased.Reliability of standard retrieval is enhanced with accuracy and precision;Hbase can effectively solve the problem of unstructured data storage in the intelligent supervision system;the construction of indexes through caching technology and Elasticsearch can greatly improve the speed of fuzzy query.