Abstract: Rapid and accurate detection of insulator defects is an important and challenging task for grid maintenance. Aiming at the problems of slow detection speed and high model complexity of the current mainstream insulator defect detection algorithms, we propose a breakage detection method of insulator based on context augmentation and feature refinement. In the method, firstly, the lightweight ECA-GhostNet is used as the backbone, and a lightweight adaptive context augmentation module is embedded at the output of the backbone to inject multi-scale context for the insulator breakage. Then, a fast and efficient feature refinement module are introduced at the output of the feature pyramid network(FPN) for augmenting insulator breakage defect features. Several sets of comparative experiments are carried out on the dataset constructed in this paper. The results show that the mean average precision(mAP) of the method proposed in this paper can reach about 97.05%, the detection speed is about 63 Frame/s, and the floating point operations(FLOPs) and parameters are 1.46 G and 1.68 M, respectively.All performance indicators are superior to those obtained by mainstream algorithms, such as RetinaNet, YOLOv4 and YOLOF. The research results can provide a reference for embedded applications of drone.