Popnet: computer vision based deep learning model for forecasting gridded population

Byeonghwa Jeong; Bo Kyeong Lee

doi:10.1080/13658816.2025.2514792

Popnet: computer vision based deep learning model for forecasting gridded population

Byeonghwa Jeong
, Bo Kyeong Lee

Human Geography

Research output: Contribution to journal › Article › peer-review

1 Scopus citations

Abstract

This study introduces Popnet, a deep learning model for forecasting 1 km-gridded populations, integrating U-Net, ConvLSTM, a Spatial Autocorrelation module and deep ensemble methods. Using spatial variables and population data from 2000 to 2020, Popnet predicts South Korea’s population trends by age groups (under 14, 15-64 and over 65) up to 2040. In validation, it outperforms traditional machine learning and state-of-the-art computer vision models. The output of this model discovered significant polarisation: population growth in urban areas, especially the capital region, and severe depopulation in rural areas. Popnet is a robust tool for offering significant insights to policymakers and related stakeholders about the detailed future population, which allows them to establish detailed, localised planning and resource allocations.

Original language	English
Pages (from-to)	217-236
Number of pages	20
Journal	International Journal of Geographical Information Science
Volume	40
Issue number	1
DOIs	https://doi.org/10.1080/13658816.2025.2514792
State	Published - 2026

Funding

This manuscript has been authored by UT-Battelle, LLC, under contract DE-AC05-00OR22725 with the US Department of Energy (DOE). The US government retains and the publisher, by accepting the article for publication, acknowledges that the US government retains a nonexclusive, paid-up, irrevocable, worldwide license to publish or reproduce the published form of this manuscript, or allow others to do so, for US government purposes. DOE will provide public access to these results of federally sponsored research in accordance with the DOE Public Access Plan (https://www.energy.gov/doe-public-access-plan).

Keywords

Gridded population
computer vision
deep ensemble
population forecasting

Access to Document

10.1080/13658816.2025.2514792

Cite this

@article{e1347a53a44245bf93cd36fb0d128391,

title = "Popnet: computer vision based deep learning model for forecasting gridded population",

abstract = "This study introduces Popnet, a deep learning model for forecasting 1 km-gridded populations, integrating U-Net, ConvLSTM, a Spatial Autocorrelation module and deep ensemble methods. Using spatial variables and population data from 2000 to 2020, Popnet predicts South Korea{\textquoteright}s population trends by age groups (under 14, 15-64 and over 65) up to 2040. In validation, it outperforms traditional machine learning and state-of-the-art computer vision models. The output of this model discovered significant polarisation: population growth in urban areas, especially the capital region, and severe depopulation in rural areas. Popnet is a robust tool for offering significant insights to policymakers and related stakeholders about the detailed future population, which allows them to establish detailed, localised planning and resource allocations.",

keywords = "Gridded population, computer vision, deep ensemble, population forecasting",

author = "Byeonghwa Jeong and Lee, \{Bo Kyeong\}",

note = "Publisher Copyright: {\textcopyright} 2025 Informa UK Limited, trading as Taylor \& Francis Group.",

year = "2026",

doi = "10.1080/13658816.2025.2514792",

language = "English",

volume = "40",

pages = "217--236",

journal = "International Journal of Geographical Information Science",

issn = "1365-8816",

publisher = "Taylor and Francis Group",

number = "1",

}

TY - JOUR

T1 - Popnet

T2 - computer vision based deep learning model for forecasting gridded population

AU - Jeong, Byeonghwa

AU - Lee, Bo Kyeong

PY - 2026

Y1 - 2026

N2 - This study introduces Popnet, a deep learning model for forecasting 1 km-gridded populations, integrating U-Net, ConvLSTM, a Spatial Autocorrelation module and deep ensemble methods. Using spatial variables and population data from 2000 to 2020, Popnet predicts South Korea’s population trends by age groups (under 14, 15-64 and over 65) up to 2040. In validation, it outperforms traditional machine learning and state-of-the-art computer vision models. The output of this model discovered significant polarisation: population growth in urban areas, especially the capital region, and severe depopulation in rural areas. Popnet is a robust tool for offering significant insights to policymakers and related stakeholders about the detailed future population, which allows them to establish detailed, localised planning and resource allocations.

AB - This study introduces Popnet, a deep learning model for forecasting 1 km-gridded populations, integrating U-Net, ConvLSTM, a Spatial Autocorrelation module and deep ensemble methods. Using spatial variables and population data from 2000 to 2020, Popnet predicts South Korea’s population trends by age groups (under 14, 15-64 and over 65) up to 2040. In validation, it outperforms traditional machine learning and state-of-the-art computer vision models. The output of this model discovered significant polarisation: population growth in urban areas, especially the capital region, and severe depopulation in rural areas. Popnet is a robust tool for offering significant insights to policymakers and related stakeholders about the detailed future population, which allows them to establish detailed, localised planning and resource allocations.

KW - Gridded population

KW - computer vision

KW - deep ensemble

KW - population forecasting

UR - https://www.scopus.com/pages/publications/105007418573

U2 - 10.1080/13658816.2025.2514792

DO - 10.1080/13658816.2025.2514792

M3 - Article

AN - SCOPUS:105007418573

SN - 1365-8816

VL - 40

SP - 217

EP - 236

JO - International Journal of Geographical Information Science

JF - International Journal of Geographical Information Science

IS - 1

ER -

Popnet: computer vision based deep learning model for forecasting gridded population

Abstract

Funding

Keywords

Access to Document

Other files and links

Fingerprint

Cite this