Abstract: Aiming at the problem of engineering landslide hazard identification in dense vegetation and steep terrain areas, an automatic identification idea of engineering landslide hazard by coupling change detection and deep learning was proposed.The hazard identification index system composed of image spectrum, NDVI,land use, elevation, slope and surface coverage vegetation change was constructed, and a deep learning convolutional neural network CNN algorithm was established.The application verification was carried out in Shexian County, Xingtai County and Kuancheng County of Hebei Province with dense vegetation and steep terrain, and 134 hidden dangers of engineering landslide hazards from 2016 to 2020 were automatically identified.The results of visual verification and field investigation showed that the recognition accuracy of this method was 91.9%,and the F1 score was 93.6%.This method is universal in the vast area, which can provide a new idea for the automatic identification of landslide hazards and a scientific basis for the rational planning of human engineering activities.