A summary of the methodology

The Global Landslide Detector monitors and analyzes Twitter image streams to capture landslide reports in real-time. The tweets collection is performed using multilingual keywords (see below the complete list) curated with the help of domain experts. The system uses several image processing machine learning models to analyze images and identify those showing landslides, mudslides, rockslides, etc. The system also detects duplicate images and only shows an incident’s first occurrence. The landslide tweets are automatically geo-tagged using location names mentioned in the tweets. Please refer to the following papers for more information about the models and system architecture.

1. Ferda Ofli, Muhammad Imran, Umair Qazi, Julien Roch, Catherine Pennington, Vanessa J. Banks, and Remy Bossu, Landslide Detection in Real-Time Social Media Image Streams, In: Neural Computing & Applications (2023). https://doi.org/10.1007/s00521-023-08648-0


2. Ferda Ofli, Umair Qazi, Muhammad Imran, Julien Roch, Catherine Pennington, Vanessa Banks, and Remy Bossu, A Real-time System for Detecting Landslide Reports on Social Media using Artificial Intelligence, International Conference on Web Engineering (ICWE), 2022
   


3. Catherine Pennington, Remy Bossu, Ferda Ofli, Muhammad Imran, Umair Qazi, Julien Roch, and Vanessa J. Banks A Near-Real-Time Global Landslide Incident Reporting Tool Demonstrator Using Social Media and Artificial Intelligence,International Journal of Disaster Risk Reduction, 2022