Generalizable Slum Detection from Satellite Imagery with Mixture-of-Experts

TL;DR

This paper introduces GRAM, a novel framework combining a large satellite imagery dataset and a Mixture-of-Experts model to improve generalization in slum detection across diverse regions without requiring labeled data from new locations.

Contribution

The paper presents a large-scale satellite imagery dataset and a two-phase test-time adaptation method using Mixture-of-Experts for robust, label-efficient slum segmentation across unseen regions.

Findings

GRAM outperforms state-of-the-art baselines in low-resource settings

The Mixture-of-Experts architecture captures region-specific features effectively

Prediction consistency filtering improves pseudo-label reliability

Abstract

Satellite-based slum segmentation holds significant promise in generating global estimates of urban poverty. However, the morphological heterogeneity of informal settlements presents a major challenge, hindering the ability of models trained on specific regions to generalize effectively to unseen locations. To address this, we introduce a large-scale high-resolution dataset and propose GRAM (Generalized Region-Aware Mixture-of-Experts), a two-phase test-time adaptation framework that enables robust slum segmentation without requiring labeled data from target regions. We compile a million-scale satellite imagery dataset from 12 cities across four continents for source training. Using this dataset, the model employs a Mixture-of-Experts architecture to capture region-specific slum characteristics while learning universal features through a shared backbone. During adaptation, prediction consistency across experts filters out unreliable pseudo-labels, allowing the model to generalize effectively to previously unseen regions. GRAM outperforms state-of-the-art baselines in low-resource settings such as African cities, offering a scalable and label-efficient solution for global slum mapping and data-driven urban planning.