This paper summarizes the traditional methods of remote sensing analysis,machine learning,and deep learning in recent years,and analyzes the advantages of applying deep learning algorithms in landslide and debris flow information extraction and recognition within the field of artificial intelligence.Through scheme comparison and experimental analysis,it is found that issues such as insufficient sample collection,poor data annotation quality,and limitations in the method of landslide feature extraction greatly restrict the accuracy and generalization of landslide intelligent recognition.In the future,we will continue to explore methods for optimizing multi-source data fusion and feature extraction,investigate landslide multi-task recognition methods based on transfer learning,and utilize landslide change detection techniques combined with remote sensing imagery to achieve accurate identification and dynamic analysis of landslides.
维普
万方数据