The program will be available shortly. Please check back later.
Presentation 1:
A human-machine collaborative approach measures economic development using satellite imagery
Presenter: Dr. Donghyun Ahn, Max Planck Institute for Security and Privacy
Abstract: Estimating socioeconomic development in regions with limited ground data is essential for understanding economies in data-scarce areas. To address this challenge, we developed a human-machine collaborative model that predicts grid-level economic development using publicly accessible satellite imagery combined with lightweight subjective ranking annotations, eliminating the need for direct ground data. Our model provides highly detailed economic insights into regions where data is limited, such as North Korea, highlighting notable development around the capital and areas with recent state-led projects. To demonstrate the model's robustness, we extended our analysis to five Asian countries, analyzing more than 400,000 grids to reveal economic landscapes. This method has the potential to inform sustainable development programs by providing granular, accessible economic data for hard-to-reach regions, supporting initiatives that align with sustainable development goals.
Presentation 2:
SEGSLUM: A semi-supervised learning monitors urban slums of low-income countries from the sky
Presenter: Dr. Jeasurk Yang, Max Planck Institute for Security and Privacy
Abstract : Identifying deprived urban settlements, or slums, plays a crucial role in sustainable urbanization by monitoring areas in need of targeted policy interventions. AI-powered approaches using satellite imagery enable accurate detection of such settlements without the need for costly surveys. However, developing a versatile model for slum detection across countries and timeframes remains a challenge, primarily due to the significant regional variations in their appearances. Here we present a semi-supervised learning model that detects slums across nine low- and middle-income countries leveraging very-high-resolution satellite imagery from 2014 to 2024. Slum areas have unchanged or increased since 2020 in most countries, suggesting that progress in improving slums may be slower than anticipated by official reports. We also apply our detection results to assess the effects of various slum redevelopment projects, revealing that slum destruction has led to the formation of new settlements near redevelopment sites. Our method generates a consistent and effective indicator for monitoring progress toward Sustainable Development Goal 11, which targets access to affordable housing and basic services.
In compliance with GDPR consent requirements, presentations given in a Visual Intelligence context may be recorded with the consent of the speaker. All recordings are edited to remove all faces, names and voices of other participants. Questions and comments by the audience will hence be removed and will not appear in the recording. With the freely given consent from the speaker, recorded presentation may be posted on the Visual Intelligence YouTube channel.
This seminar is open for members of the consortium. If you want to participate as a guest please sign up.
Dr. Donghyun Ahn and Dr. Jeasurk Yang, Max Planck Institute for Security and Privacy
This seminar is open for members of the consortium. If you want to participate as a guest please sign up.
Dr. Donghyun Ahn and Dr. Jeasurk Yang, Max Planck Institute for Security and Privacy
This seminar is open for members of the consortium. If you want to participate as a guest please sign up.