The main landslide detection methods based on remote sensing include (1) Pixel-based, (2) Object-oriented, (3) Machine Learning, and (4) Deep Learning. As a consequence, how to detect landslides rapidly in large-scale remote sensing datasets is of great significance for disaster mitigation. The development of advanced earth observation techniques provides unique opportunities for the comprehensive assessment of disaster losses, thus massive amounts of satellite images are gradually replacing field surveys as a low-cost resource for building a landslide database. Therefore, generating landslide inventory is the first step in hazards management and susceptibility evaluation. Additionally, studies have shown that the legacy effect may lead to a higher chance of landslides occurring on previous landslide paths in the next decade. A detailed landslide inventory records the location, date, and type of landslide in a given area, which can serve as a basis for further investigation and provide geologists with scientific data used to assess landslide risk. Once landslides develop into geological hazards, there is a great potential to cause devastating damage to natural structures and infrastructure, leading to human casualties and property damage. The results indicated that the proposed model obtained the highest mIoU and F1-score in both datasets, demonstrating that the ResU-Net with a transformer embedded can be used as a robust landslide detection method and thus realize the generation of accurate regional landslide inventory and emergency rescue.Īs one of the most critical types of natural hazards, landslides are triggered by various external factors in most cases, including earthquakes, rainfall, variation of water level storms, and river erosion. Finally, the standard ResU-Net was chosen as the benchmark to evaluate the proposed model rationality. By selecting two landslide datasets with different geological characteristics, the feasibility of the proposed model was validated. Besides, a spatial and channel attention module was introduced into the decoder to effectually suppress the noise in the feature maps from the convolution and transformer. Inspired by the vision transformer (ViT), this paper first attempts to integrate a transformer into ResU-Net for landslide detection tasks with small datasets, aiming to enhance the network ability in modelling the global context of feature maps and drive the model to recognize landslides with a small dataset. How to effectively integrate transformers into CNN, alleviate the limitation of the receptive field, and improve the model generation are hot topics in remote sensing image processing based on deep learning (DL). Currently, the success of transformers in natural language processing (NLP) demonstrates the strength of self-attention in global semantic information acquisition. Compared to a traditional landslide detection approach, convolutional neural networks (CNN) have been proven to have powerful capabilities in reducing the time consumed for selecting the appropriate features for landslides. An efficient method of landslide detection can provide basic scientific data for emergency command and landslide susceptibility mapping.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |