Electrical Equipment Classification via Improved Faster Region-based Convolutional Neural Network

In this paper, an improved faster region-based convolutional neural network (Faster R-CNN) model is proposed to identify electrical equipment in complex background or small target images. The Batch Normalization (BN) layer is added to the residual network of improved Faster R-CNN for feature extraction. The improved Faster R-CNN in the residual stage deepens the network depth and obtains richer feature information while keeping the number of model parameters unchanged. The improved network model enhances the ability of deep semantic information retention and anti-interference, thus improving the detection accuracy. The simulated results show that the improved network model effectively improves the recognition speed and classification accuracy. The accuracy of the model reaches 95%, and the processing speed reaches 12ms per image.