Automatic detection method for damage of glass curtain wall of high-rise building

Abstract

The invention relates to an automatic detection method for damage of a glass curtain wall of a high-rise building, and the method comprises the steps: constructing a detection network, carrying out the training and verification of the constructed detection network through a sample set, and obtaining a detection network with optimal parameters. A convolutional layer in the YOLO v4 network is replaced by expansion convolution, channel-by-channel convolution and point-by-point convolution which are connected in sequence to obtain a new convolutional layer, and finally, the YOLO v4 network after replacement is used as a constructed detection network; pruning the detection network with the optimal parameters according to the scaling factor vectors of all the BN layers; and finally, performing fine tuning on the pruned detection network with the optimal parameters by using the sample set to obtain a final detection network. Therefore, the method further reduces the parameter quantity of the model on the premise of ensuring high accuracy, and realizes the full-automatic detection of the glass curtain wall of the urban high-rise building based on the unmanned aerial vehicle.

Description

technical field [0001] The invention relates to the field of image detection, in particular to an automatic detection method for damage to glass curtain walls of high-rise buildings. Background technique [0002] Architectural glass curtain wall is widely used in high-rise buildings because of its beautiful appearance, wide field of view, quick construction and strong plasticity. In the architectural glass curtain wall industry, my country has experienced rapid development from scratch to a dominant position in the market, and has now become the world's largest producer and user. However, affected by external factors (such as bad weather), use time and other factors, the glass curtain wall will be damaged during use, which not only affects the beauty and vision of the glass curtain wall, but also poses a huge safety hazard, so it must be Strengthen the timely and accurate detection of building glass curtain walls. For this reason, my country has promulgated a strict archit...

AI Technical Summary

Problems solved by technology

In addition, the feature extraction backbone network and feature fusion module contain multiple convolutional layers, and a BN layer (batch normalization layer) is installed after each convolutional layer; however, the YOLO v4 network has a large number of parameters, which is difficult to embed into Drones, small devices with limited computing power

Images