Gave my first invited talk at the 6th International Electronic Conference on Remote Sensing (ECRS 2026)

On 21 October 2026 I gave my first invited talk, at the 6th International Electronic Conference on Remote Sensing (ECRS 2026), Global Perspectives for the Next-Generation in Earth Observation, hosted online by MDPI and Remote Sensing. I presented in Session S5, Remote Sensing in Urban Areas, chaired by Prof. Dr. Riccardo Buccolieri, in the invited slot of Stefanos Georganos, on his behalf. The session also featured an invited talk by Prof. Dr. Monika Kuffer on urban nights from space.

Hidden deficits: mapping urban deprivation across 5,000+ cities with Earth observation and GeoAI

The talk was based on our Nature Cities paper. More than a billion people live in slums and informal settlements, yet household surveys and censuses rarely show where deprivation clusters within a city. Urban form can fill that gap: building size and spacing, density and street access are visible from above, in every city.

We classified neighbourhood-scale city segments in 5,132 cities across 103 countries in Africa, Asia and Latin America and the Caribbean. We estimate that about 395 million people (20.2% of the urban population covered) live in morphologically deprived segments, and that more than a third of them, around 136 million, live in small and medium-sized cities, not the megacities that usually get the attention.

Over one-third live in small and medium-sized cities

Three small cities, three different patterns

The map is meant as a starting point for local investigation: a way to screen and prioritise areas alongside local assessment, not a substitute for it. The data are open on Zenodo, the code is on GitHub, and you can explore your own city in the CSMD viewer.

I closed with two ongoing studies that build on this work: how many people live in these areas, and whether satellite embeddings are good enough as an input for mapping them.

Thank you to Stefanos for trusting me with the talk, to the session chair and the ECRS 2026 organisers, and to everyone who joined online. The work is funded by FORMAS through DEPRIMAP (2023-01210), with reference data from IDEAtlas and IDEABench.