[TMP-065] Intelligent crowdsourcing of geolocation tasks in natural disasters
The leveraging of crowdsourcing and artificial intelligence to help emergency responders build damage maps in disaster-hit areas
Social media generates vast amounts of real-time data, which is crucial during the first 72 hours after a disaster. While there are advanced machine learning methods for classifying and geolocating social media images, the challenge of efficiently using this data for disaster response remains unsolved.
Currently, the typical approach involves filtering images and then randomly assigning them to a crowd of volunteers for geolocation. This project aims to improve this process by leveraging crowdsourcing and artificial intelligence to help emergency responders build damage maps in disaster-hit areas. The platform will intelligently assign geolocation tasks to volunteers based on their skills, learned from their previous geolocation experiences.
The project focuses on two main tasks:
1. Profile Learning: Develop volunteer profiles based on their past geolocations, representing their ability to accurately geolocate images.
2. Active Task Assignment: Use these profiles to efficiently distribute tasks, ensuring high geolocation quality and fair task distribution.
The first stage will involve testing with artificial data as a feasibility study, with plans to integrate the system with the crowdnalysis library. Future work includes a geolocation workshop in Barcelona and the development of a platform that will help organizations quickly assess disaster damage, improving emergency responses.
The project focused on improving the accuracy and efficiency of geolocating social media images during emergencies by using crowdsourced volunteers. Key results include the development of two models: a profile-learning model to gauge volunteers’ geolocation abilities and a task assignment model that optimizes image distribution based on volunteer skills. These models outperform traditional random assignment approaches by reducing annotation requirements and improving the quality of geolocation consensus without sacrificing accuracy. This method holds promise for disaster response applications. We had 3 main outputs:
- Open-source implementation of the volunteer profiling and consensus geolocation algorithms into the crowd analysis library.
- Papers with the evaluation of the different geolocation consensus and active strategies for geolocation:
- an online workshop to collect expert feedback about the topic
Tangible Outcomes
- Ballester, Rocco, Yanis Labeyrie, Mehmet Oguz Mulayim, Jose Luis Fernandez-Marquez, and Jesus Cerquides. “Crowdsourced Geolocation: Detailed Exploration of Mathematical and Computational Modeling Approaches.” Cognitive Systems Research 88 (December 1, 2024): 101266. https://doi.org/10.1016/j.cogsys.2024.101266 .
- Cerquides J., Mülâyim M.O. Crowdnalysis: A software library to help analyze crowdsourcing results (2024), 10.5281/zenodo.5898579 https://github.com/IIIA-ML/crowdsourced_geolocation
Partners
- Consejo Superior de Investigaciones Científicas (CSIC), Jesus Cerquides, cerquide@iiia.csic.es
- University of Geneva, Jose Luis Fernandez Marquez