PROBES snapshot: MarketVision, computer vision for market monitoring

This one-page brief, developed by the Digital Transformation Accelerator (DTA) team at the Alliance of Bioversity International and CIAT, is based on the MarketVision probe led by Thibaud Vantalon, Deep learning specialist at the Alliance Bioversity–CIAT. The probe explores whether smartphone videos combined with computer vision and 3D mapping can be used to monitor informal wet markets, where food availability, prices, and diversity are rarely captured through conventional data systems. By processing short walkthrough videos, the approach aims to detect food items, estimate quantities, identify prices, and map stall layouts in highly dynamic market environments. Implemented as an early-stage, safe-to-fail experiment, the probe focuses on learning how well retail-analytics AI performs in informal settings and what ethical, technical, and operational constraints must be addressed. The brief synthesizes early insights to inform future development of scalable tools for urban food system monitoring, nutrition analysis, and market resilience.