In order to obtain the development rules of ground cracks in mining areas timely and accurately, monitor the mining surface damage and restore the ecological environment of mining areas, the mining face of Gaojialiang Coal Mine in Wanli Mining area of Inner Mongolia Province was taken as the research area, and the identification and development rules of ground cracks in mining areas were studied based on UAV remote sensing technology and deep learning algorithm. First, ArcGIS software was used to crop the study area in batches, and more than 10,000 small images were obtained. Secondly, we use the classification network model to classify small images with or without cracks. On this basis, semantic segmentation model is used to identify crack images, and crack development rules are analyzed through image Mosaic. The UAV remote sensing technology and deep learning algorithm can effectively extract surface cracks in mining areas, and the research can provide effective support for controlling cracks, preventing disasters and restoring ecological environment in mining areas.