Low-Cost Biotic and Abiotic Stress Prediction Using Artificial intelligence

David Hughes of Pennsylvania State University in the U.S. is leveraging real-time, high-resolution satellite imagery of smallholder farms along with artificial intelligence to automatically detect crop pests and diseases in Africa. In Phase I, together with Nita Bharti also of Penn State University in the U.S. and James Legg at the International Institute of Tropical Agriculture in Tanzania and the Charity, Self Help Africa, they recruited 10 agricultural graduates to collect detailed field data on crop pests across Busia County in Kenya together with a large network of farmers supplied with smartphones carrying an artificial intelligence-based visual diagnostic tool. They also trained machine learning models using satellite data to accurately predict future crop growth. In Phase II, they will combine different remote-sensing observations from satellites and other relevant data, such as soil quality and water stress, with crowdsourced observations from 400,000 farmers across Kenya together with machine learning models to build a low-cost sentinel surveillance system with enhanced accuracy. They will test the ability of this system to detect the top five diseases for the top five crops and to provide accurate pest forecasts that can help smallholder African farmers grow more food and make greater profits.