Insulators are important devices for circuit insulation and fixing wires in transmission lines, and their surfaces are often damaged by lightning discharge, rainwater, and other harsh natural environments. To address the issues of low recognition accuracy, complex recognition background, and small targets caused by outdoor lighting and complex background, this paper proposes a transmission line insulator defect detection method based on improved YOLOv7. The model introduces the CNeB module, which uses convolutional kernels of different sizes and depths to capture features at different scales, thereby helping to capture the diversity and complexity of the target. Secondly, the GAM attention mechanism is introduced in the head part to reduce information dispersion while enhancing the important features of the target in both spatial and channel dimensions. Finally, the Wise IOU loss function is used to improve the accuracy and generalization ability of the model. Through testing on the public dataset CPLID, the average accuracy value of this model is 98.3%, which is 5% higher than the Yolov7 model, fully proving that this model can meet the requirements of insulator defect detection.