Game-changers in global slum detection? A review of remote sensing for assessing complexity of informal settlements

Informal settlements, now home to over one billion people globally, pose significant challenges for sustainable urban development. Effective monitoring and management of these dynamic environments are impeded by the traditional data collection methods like ground surveys and census enumeration. Remote sensing (RS), augmented by advancements in artificial intelligence, offers a scalable and timely solution for identifying and monitoring these settlements. This study provides a systematic literature review of 53 peer-reviewed articles published between 2000 and 2024 (a sample which was extracted from initial 635 articles identified through the database search), analysing the application of RS technologies in this domain. Our analysis reveals several key trends: a strong geographic focus on the Global South, particularly Asia and Africa; a methodological evolution from traditional pixel-based and object-based image analysis (OBIA) to a current predominance of machine learning and deep learning (DL) techniques, which now account for 43% of the reviewed studies. While modern DL approaches frequently report high overall classification accuracies ranging from 80% to 95%, our analysis reveals a persistent deficiency in the standardization and consistency of accuracy assessment reporting across the literature and a diverse utilization of satellite data, with medium-resolution archives like Landsat and Sentinel used for large-area monitoring and very-high-resolution (VHR) commercial imagery employed for detailed morphological analysis. While advanced DL models have improved detection accuracy, significant challenges persist, including the limited transferability of models across different geographic contexts and the difficulty in capturing socio-economic dimensions from physical features alone. This review synthesizes the current state of the field: it identifies critical research gaps, and proposes future directions, emphasizing the need for context-specific methodologies, standardized validation protocols, and the integration of multi-source data to create more holistic and equitable monitoring frameworks for informal settlements worldwide.