Transmission Line Image Defect Diagnosis Preprocessed Parallel Method Based on Deep Learning

Abstract

Deep learning has shown its potential and advantages in feature extraction and model fitting. It has also been promoted and applied in the field of transmission, and cooperates with drone inspection data to diagnose defects of transmission equipment. This paper aims at engineering application. The accuracy of defect diagnosis is improved, and multiple significant target detection is applied. Preprocess the image and implement parallel image prediction. The image preprocessing includes defogging processing, rotation, etc., and finally achieves target detection by combining logic judgment.