Unmanned aerial vehicles (UAVs) equipped with HD camera can obtain a large number of detailed inspection images of the insulators which are indispensable components in the transmission lines. A quick and accurate detection of the insulator can provide a reliable basis for distance measurement and obstacle avoidance for UAV when flying close to towers for details. Simultaneously, as a faultprone component, the insulators seriously threat the network security, thus computer technology is required for fault diagnosis. The detection of the insulator image with the complex aerial background is implemented by constructing a convolutional neural network (CNN), which has the classic architecture of five modules of convolution and pooling, two modules of fully connected layers. In this paper, a recognition algorithm for explosion fault based on saliency detection is proposed, which uses the trained network model to extract the features. Then we put the saliency maps into a SOM network and build the mathematical module via super pixel segmentation, contour detection, and other image processing methods. The test shows that the algorithm can reduce the errors caused by manual analysis. The test also demonstrates that the detection rate of the insulator can reach 90% and the recognition rate of explosion fault can reach 85% with complex background. They effectively improve the efficiency of inspection and make the inspection more intelligent.