A Light Weight Mobile Net SSD Algorithm based dentification and Detection of Multiple Defects in Ceramic Insulators

Presently, transmission lines play a major role for transferring electricity from generating stations to consumer premises. As the components are exposed to environmental conditions, damage to these components is more severe. Among those components insulators area accountable for 81.3% of line interruptions. In this paper, a drone-based transmission line inspection using mobile net single shot detection (SSD) is proposed that aims to detect the potential defects of insulators. The proposed system employs a regression-based algorithm using mobile net uses an upgraded compact mobile net SSD to identify targets network, allowing for precise categorization and placement of electrical equipment. Simulation is performed by loading the datasets that are collected. There are totally 800 images of ceramic insulators these are sub divided into 3 classes train class has images with 590 test class with 130 and validation (Val) class with 80. In this methodology, mobile net SSD can identify and detect the ceramic insulators with pollution flashover, corroded and broken. Precision, recall, F1 Score are calculated. Analysis is carried out on the variation of batch parameters, size etc to ascertain the effectiveness of the proposed method.