The safe operation of transmission lines is often threatened by hidden dangers in line corridors such as illegal buildings, tree barriers and large illegal construction vehicles. Aiming at the problems of limited working conditions, small monitoring range and low accuracy of current line corridor hidden danger detection methods, this paper presents an improved line corridor hidden danger detection method based on U-net semantics segmentation network, trains the model based on a small amount of satellite remote sensing image data, and achieves fast and accurate detection of hidden dangers in line corridors, the detection accuracy is 85%. The experimental results show that the proposed method has a good detection effect for corridor hidden dangers and improve the level of intelligent detection of hidden dangers in transmission line corridors.