Loess Landslide Detection Using Object Detection Algorithms in Northwest China

Weile Li,Yuanzhen Ju,Shichao Jin,Xiujun Dong,Qinghua Guo,Yanjun Su,Qiang Xu

doi:10.3390/rs14051182

Abstract

Regional landslide identification is important for the risk management of landslide hazards. The traditional methods of regional landslide identification were mainly conducted by a human being. In previous studies, automatic landslide recognition mainly focused on new landslides distinct from the environment induced by rainfall or earthquake, using the image classification method and semantic segmentation method of deep learning. However, there is a lack of research on the automatic recognition of old loess landslides, which are difficult to distinguish from the environment. Therefore, this study uses the object detection method of deep learning to identify old loess landslides with Google Earth images. At first, a database of loess historical landslide samples was established for deep learning based on Google Earth images. A total of 6111 landslides were interpreted in three landslide areas in Gansu Province, China. Second, three object detection algorithms including the one-stage algorithm RetinaNet and YOLO v3 and the two-stage algorithm Mask R-CNN, were chosen for automatic landslide identification. Mask R-CNN achieved the greatest accuracy, with an AP of 18.9% and F1-score of 55.31%. Among the three landslide areas, the order of identification accuracy from high to low was Site 1, Site 2, and Site 3, with the F1-scores of 62.05%, 61.04% and 50.88%, respectively, which were positively related to their recognition difficulty. The research results proved that the object detection method can be employed for the automatic identification of loess landslides based on Google Earth images.

Highlights

When F1-scores were used to evaluate the accuracy of the fusion result on Mask R-Convolution neural network (CNN), the F1-score was 55.31%, and the corresponding precision and recall were 47.41% and 66.37%, respectively
F1‐Score between the three models is that YOLO v3 and RetinaNet belong to a one-stage algorithm, 57.42%
A sample set of historical loess landslides was first constructed for deep learning

Summary

Introduction

China has a vast territory, variable terrain, and frequent geological hazards that seriously threaten people’s lives, especially landslides. Landslide identification is the basis for other research works. The identification of landslides can be divided into two categories. One is the identification of an indication before the landslide. Methods in this category are to observe the deformation of the slope based on multitemporal data. The object of the observation is the unstable slope that has not yet become a disaster [1], and the research methods include InSAR technology [2] and multi-period terrain data [3]. The other category is the identification of landslides that have already occurred. The results of landslide detection can be used for the study of Remote Sens.

Objectives

Results

Discussion

Conclusion