On the power of CNNs to detect slums in Brazil

Bibliographic Data

ID7150599
AuthorsJoão Pereira Da Silva (0000-0001-7328-9974, Institute of Mathematics and Computer Science, corresponding author), José F Rodrigues-Jr, José F Rodrigues (0000-0001-8318-1780, Institute of Physics), João Porto De Albuquerque (0000-0002-3160-3168, Urban Big Data Centre)
Year2025
Volume121
Pages102306
Publication date2025-10-01
Peer ReviewedYes
Open AccessYes
TypeARTICLE
VenueComputers Environment and Urban Systems (JOURNAL)
Journal identifiersISSN: 0198-9715 • E-ISSN: 1873-7587
PublisherElsevier BV (PUBLISHER)
DOI10.1016/j.compenvurbsys.2025.102306
OpenAlexW4410911680
LanguageEN
References cited27

The rapid expansion of slums poses a critical challenge for urban planning in Low- and Middle-Income Countries (LMICs), where traditional data collection methods like censuses are often outdated and insufficient. This study examines the transferability and generalization capabilities of deep learning models, specifically Convolutional Neural Networks (CNNs), for automated slum detection across six Brazilian cities with varying urban morphologies: São Paulo, Rio de Janeiro, Belo Horizonte, Brasília, Salvador, and Porto Alegre. Utilizing Very High Resolution (VHR) and High Resolution (HR) satellite imagery, we trained and evaluated models based on the EfficientNetV2L architecture. Our experimental results show that CNN models trained on data from a single city achieved high accuracy within that city (F1 scores exceeding 0.90 with VHR imagery), but their performance significantly decreased when applied to other cities (F1 scores dropping below 0.80), highlighting the impact of regional variations in urban morphology. Conversely, a generalized model trained on combined data from all six cities maintained robust performance across all cities, achieving F1 scores above 0.80 with VHR imagery. These findings indicate that while CNNs are effective for automated slum mapping, regional diversity necessitates training on diverse datasets to ensure generalization. We provide a comprehensive methodology over an openly shared dataset, and code to facilitate future research and applications in urban geoscience. The aim is to enhance the scalability and generalization of remote sensing and deep learning methods for slum identification across diverse urban environments. • Deep learning models can accurately detect slums in diverse Brazilian cities. • Generalized models exhibit robustness across different urban landscapes. • Regional variations require further refinement for optimal model generalization. • Comprehensive guidance on rich Brazilian datasets is provided to support future research. • Very High Resolution imagery significantly outperforms High Resolution imagery

Cartography · Geography · Human Mobility and Location-Based Analysis · Land Use and Ecosystem Services · Urban and Rural Development Challenges

Tools

Open DOIOpen Access