This paper proposes a lightweight insulator defect detection model based on the Faster R-CNN-tiny architecture to address the issues of high parameter count, low accuracy, and slow detection speed in traditional object detection methods for insulator defects. The model replaces the ResNet backbone with EfficientNet, which reduces model parameters while improving detection accuracy. A feature pyramid network is used to generate multi-resolution feature maps for better detection of objects at various scales. Depth-wise separable convolutions are introduced to enhance detection speed while slightly reducing accuracy. Transfer learning is applied to improve the model's ability to detect small target defects. The model is validated using a self-exploding insulator defect dataset, showing significant improvements in mean average precision (mAP), frames per second (FPS), and parameter count compared to the original Faster R-CNN model. The proposed model effectively detects insulator defects and is suitable for edge deployment. The research highlights the importance of lightweight detection models in power system inspections and demonstrates the effectiveness of the proposed method in improving detection accuracy and efficiency.This paper proposes a lightweight insulator defect detection model based on the Faster R-CNN-tiny architecture to address the issues of high parameter count, low accuracy, and slow detection speed in traditional object detection methods for insulator defects. The model replaces the ResNet backbone with EfficientNet, which reduces model parameters while improving detection accuracy. A feature pyramid network is used to generate multi-resolution feature maps for better detection of objects at various scales. Depth-wise separable convolutions are introduced to enhance detection speed while slightly reducing accuracy. Transfer learning is applied to improve the model's ability to detect small target defects. The model is validated using a self-exploding insulator defect dataset, showing significant improvements in mean average precision (mAP), frames per second (FPS), and parameter count compared to the original Faster R-CNN model. The proposed model effectively detects insulator defects and is suitable for edge deployment. The research highlights the importance of lightweight detection models in power system inspections and demonstrates the effectiveness of the proposed method in improving detection accuracy and efficiency.