dataset of transmission line insulator

This data set collects aerial images of transmission line insulators and their defects by UAV aerial photography, and carries out sample collection for real grid inspection scenes. The image capture process fully restores the complex environment of field inspection, and the sample background covers a variety of real scenes such as fields, mountains, forests, sky, buildings, etc., including actual working conditions such as light changes, complex background interference, target scale differences, etc. The data set not only contains normal insulator samples, but also covers a variety of typical insulator defect types such as self explosion, damage, and contamination. The sample background and defect types are rich and diverse, which can fully simulate all kinds of interference factors encountered in the process of UAV patrol inspection, and meet the actual engineering requirements of transmission line insulator defect detection, and can be used for the training, verification, and performance evaluation of target detection model. This data set has a total of 6400 images. The number of samples is sufficient. Considering the distribution of positive and negative samples, it can provide sufficient data support for the deep learning model to learn insulator target characteristics and distinguish normal samples from defective samples. It is suitable for the research of insulator defect detection algorithm for edge deployment.