Landslide is a kind of natural disaster that occurs frequently in our country, which is often extremely destructive and hazardous, and poses a serious threat to people's life and property safety and the ecological environment. Landslide identification is important for post-disaster rescue and post-disaster assessment. The traditional landslide identification method has the problems of low identification efficiency and strong subjectivity. In this paper, an improved Faster-RCNN landslide identification method is proposed. Based on the Faster-RCNN model, the method replaces the original backbone network VGG-16 with ResNet-50, which has more excellent feature extraction capability, and introduces the region of interest alignment (ROI Align) strategy to improve the model's candidate frame localization accuracy, so as to further enhance the overall detection effect of the model. The experimental results show that the improved Faster-RCNN model improves both accuracy and recall, and is able to quickly and accurately detect landslides in complex backgrounds, which has high application value.