Fault detection method based on residual network and Faster R-CNN

Abstract

To improve the fault detection accuracy, a method based on residual network and Faster R-CNN is proposed. First, input the image into the ResNet-50 feature extraction network to obtain the corresponding feature map, and then use the RPN structure to generate a candidate frame, and project the candidate frame generated by the RPN onto the feature map to obtain the corresponding feature matrix. Finally, each feature matrix is scaled to a fixed-size feature map through the ROI pooling layer, and then the feature map is flattened through a series of fully connected layers to obtain the prediction result. ResNet50 solves the problem of network degradation and over-fitting caused by the deepening of network layers when extracting features from faults. Faster R-CNN implements end-to-end training, combining the advantages of ResNet50 and Faster-RCNN, and has accurate positioning efficiency. In the aspect of data enhancement, it is further optimized to enhance the generalization ability of the network, optimize the detection results of the network, and effectively improve the accuracy of the verification, and the feasibility of the method is verified through actual seismic data.