Transmission line is particularly important as a regular inspection of China's power transmission network. The traditional manual inspection method of transmission lines has been gradually eliminated with the development and application of UAV automatic inspection technology due to its low efficiency. Therefore, this study systematically studies the detection of UAV inspection defects in transmission lines based on deep convolutional neural networks. Firstly, the typical defect set of transmission line UAV inspection is constructed. Secondly, based on the deep convolutional neural network, a typical defect detection model for UAV inspection of transmission lines is constructed and trained. Finally, the typical defect detection experiment of transmission line UAV inspection is carried out and analyzed. This study realizes a high-precision and low-cost automatic detection network for typical defects of transmission lines.