---

FieldValAI: analysis ready AI training data for smallholder climate change adaptation strategies

PI and co-PIs: Berber Kramer (International Food Policy Research Institute, Kenya); Koen Hufkens (BlueGreen Labs, Belgium); Lilian Waithaka (ACRE Africa, Kenya); Senthilkumar Sankarraju (Dvara E-Registry, India)

Funding amount: $124,500.00

Project overview: More frequent and unpredictable extreme weather events are causing highly variable crop losses, putting the livelihoods of smallholder farmers across Africa and Asia at increasing risk. Efforts to strengthen farmers’ adaptive capacity are limited by the absence of large, systematic, field‑validated data on crop growth, disturbances, and management practices - data essential for training, validating, and improving climate‑relevant agricultural analytics and AI models. FieldValAI addresses this gap by creating an openly accessible, standardized dataset linking farmer‑generated images of crop growth and disturbances with matched satellite, climate, and environmental data to support machine learning applications. This dataset is developed in collaboration with smallholder‑focused social enterprises in India and Kenya that provide credit, insurance, and advisory services, ensuring the data reflect real operational needs and can inform future improvements in these services. By releasing this dataset as a public good and sharing workflows through open dissemination, hands‑on training, and outreach at major AI‑for‑agriculture convenings in South Asia and Africa, the project enables future research, tool development, and broader geographic scaling. Ultimately, FieldValAI supports innovation toward more climate‑resilient food systems and enhances long‑term adaptation capacity across the smallholder agriculture sector.

Climate Finance & Economics Agriculture & Food Earth Observation & Monitoring Societal Adaptation & Resilience Computer Vision & Remote Sensing