Automatic segmentation of landslide scars in remote sensing imagery with deep learning

• Grant number: 19/17555-1
• Support Opportunities: Scholarships in Brazil - Master
• Start date: December 01, 2019
• End date: November 30, 2021
• Field of knowledge: Physical Sciences and Mathematics - Geosciences - Geology
• Principal Investigator: Carlos Henrique Grohmann de Carvalho
• Grantee: Lucas Pedrosa Soares
• Host Institution: Instituto de Geociências (IGC). Universidade de São Paulo (USP). São Paulo , SP, Brazil

Abstract

The global number of landslides cases has increased due to greater urbanization and land use, as well as higher frequency of extreme climatic events. Since these events have strong social and economic impacts on Brazil and worldwide, they are considered one obstacle for the sustainable development according to the United Nations (UN). Studies conducted to date aim to find landslide prone areas, nevertheless, few studies address automatic methodologies for systematic monitoring of these areas. Thus, the main objective of this study is to use deep learning with convolution neural networks (CNN) in order to automate the segmentation of landslide scars on remote sensing imagery. The automation of this process is relevant to reduce the social and economic impacts since it allows constant monitoring of landslide susceptible areas, moreover, it is important for the validation of predictive models. The methodology that will be used, aiming to achieve the expected results, addresses aspects related to the database, architecture, hyperparameters, validation and result metrics of the model. The CNN will be constructed with the Python libraries: Keras and TensorFlow. (AU)

Scientific publications

(References retrieved automatically from Web of Science and SciELO through information on FAPESP grants and their corresponding numbers as mentioned in the publications by the authors)

XU, GUOSEN; WANG, YI; WANG, LIZHE; SOARES, LUCAS PEDROSA; GROHMANN, CARLOS H.. Feature-Based Constraint Deep CNN Method for Mapping Rainfall-Induced Landslides in Remote Regions With Mountainous Terrain: An Application to Brazil. IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING, v. 15, p. 16-pg., 2022-01-01. (19/17555-1, 16/06628-0)

Academic Publications

(References retrieved automatically from State of São Paulo Research Institutions)

SOARES, Lucas Pedrosa. Segmentação automática de cicatrizes de deslizamento de terra em imagens de sensores remotos utilizando aprendizagem profunda de máquina (Deep Learning). 2022. Master's Dissertation - Universidade de São Paulo (USP). Instituto de Geociências (IG/BT) São Paulo.