A new research paper evaluates the necessity of sub-meter resolution imagery for accurate cocoa mapping in Cote d'Ivoire. The study found that very high resolution (VHR) imagery, specifically 0.5 m Pleiades, achieved the highest performance (F1 = 0.92) and maintained accuracy across various landscape conditions. While decametric inputs like TESSERA (F1 = 0.86) and foundation-model embeddings from AlphaEarth Foundations (AEF) (F1 = 0.82) offer scalable alternatives, VHR imagery proved particularly beneficial in complex, fragmented landscapes. AI
IMPACT Foundation models offer a scalable alternative for large-area cocoa mapping, potentially aiding deforestation monitoring and supply-chain transparency.
RANK_REASON The cluster contains a research paper published on arXiv detailing a study on Earth observation imagery for cocoa mapping.
AI-generated summary · Google Gemini · from 2 sources. How we write summaries →
