Insulator Fault Recognition Based on Spatial Pyramid Pooling Networks with Transfer Learning (Match 2018)

Abstract

The insulators are the key component in the power transmission systems. In general, the images of insulators are difficult to obtain and the size of it may variance due to shooting angle and distance. The size of images has a great importance in computer vision and image processing. In this paper, a method of Insulator Fault Recognition Based on Spatial Pyramid Pooling networks (SPP-Net) with transfer learning is proposed to process the dataset which is small and the size of images are variance. The proposed method mainly employs the SPP-Net and with a transfer learning. The SPP-Net is used to relieve the constraint of fixed-size of the Deep Convolutional Neural Networks and learn the information from ImageNet Large Scale Visual Recognition Challenge (ILSVRC). Then the method of transfer learning is used to transfer the information learned by SPP-Net to the new model which will be used to the small dataset of insulators to improve the detection of the status of insulators. This paper conduct a thorough evaluation experiment on a real image-dataset of insulators. The experimental results indicate that the proposed method is suitable for this kind of insulators.