Abstract: For land use planning, management/assessment, geodisaster risk mitigation, as well as post-disaster reconstructions, accurate landslip detection and mapping is crucial. The most common methods for mapping landslides up to this point have been visual interpretation and field survey. These methods are frequently criticised for being labor-intensive, time-consuming, and expensive. The deep-learning-based strategy for landslip detection and mapping has received a lot of interest due to its major benefits over the conventional techniques in light of the quick development of artificial intelligence. However, the use of a deep-learning-based approach [1] in landslip identification from satellite photos has long been limited by a lack of sufficient training samples. Studies comparing the suggested approach's viability and robustness to those of ResUNet and DeepUNet showed that it has significant potential for use in the emergency response to natural catastrophes. H5 keras model was developed and adopted. We have also considered earthquake dataset all over the world and with the help of cloud computing the impact of disaster by earthquake will be predicted.
Keywords: Cloud computing, heroku cloud, Multichannel output with cascading, H5 Convolutional Neural Network model, Convolution Neural Network (CNN) architecture, geodisaster, earthquake, landslide, ResUNet and DeepUNet.
|
DOI:
10.17148/IJARCCE.2023.12605
