The Promise of Combining Satellite Imagery and Deep Learning to Predict Changes in Population at IGARSS 2022
Vinayaraj Poliyapram, Data Scientist at RIT, presented his latest research paper at this year’s International Geoscience and Remote Sensing Symposium (IGARSS). Vinay’s work, “A MULTI-TASK DEEP LEARNING MODEL FOR POPULATION AND LULC (M2PL-NET) PREDICTION WITH SCALING TO A PEOPLE FLOW GRID,” was written in collaboration with Jeremiah Anderson and Mayank Bansai of Rakuten Group, Inc.
Population and land use and land cover (LULC) projections can be critical for decision-making in urban planning, disaster risk management, oil production, health and disease containment, supply chain activities, and several other areas. Deep learning has been applied to satellite data, which has been transformative in helping us better understand population landscapes and estimations. However, many challenges remain including missing data points in rural areas and in difficult terrain.
Vinay’s paper proposes a new approach to address challenges by combining a multi-task deep learning satellite imagery technique with user GPS trajectories to predict population. He finds that his model can be an effective tool for prediction forecasts, LULC projections, and various use-cases.
The International Geoscience and Remote Sensing Symposium (IGARSS) is the flagship conference of the IEEE Geoscience and Remote Sensing Society (GRSS). It is aimed at providing a platform for sharing knowledge and experience on recent developments and advancements in geoscience and remote sensing technologies, particularly in the context of earth observation, disaster monitoring and risk assessment.
IGARSS 2022 was held in Kuala Lumpur, Malaysia from July 17-22.