Design of the Safety Monitoring System for Civil Engineering Construction

Abstract

In order to promote the development of automation, informatization and intelligence of civil engineering safety management, this paper proposes the framework of intelligent discovery and abnormal detection strategy in surveillance video. Based on the analysis of the characteristics and requirements of project monitoring, we put forward the abnormal event discovery technology of construction video monitoring. Then, SVM+CNN model are respectively used for image classification and feature extraction of risk recognition. At the same time, the adaptive pooling layer is introduced to filter the discriminant information during the training process. The case study is under the real environment of civil engineering construction. The test results show that our strategy can effectively identify abnormal events in construction monitoring, and it shows better comprehensive performance compared with similar algorithms.