Abstract: To solve the problem of poor detection results caused by different sizes of insulator defects in inspection images, a multi-scale context aware defect detection network (CAD²Net) is proposed. The network adopts ResNeSt101 architecture to improve the feature extraction capability of images. An improved feature pyramid structure is designed to detect targets at different scales by rich semantic feature maps with different resolutions. At the same time, the adaptive receptive field (Ada-RF) module is added to the detection unit of the network to aggregate multi-scale context information, generate more discriminative features, and improve the detection effect of the network on targets of different scales. The average detection accuracy of the randomly generated defects dataset and Chinese power line insulator dataset is 91.7% and 91.0% respectively. The results show that the proposed defect detection network can accurately identify and locate the defects of insulators of different scales.