Predicting the Freshness of fruits on store level by satellite images of the growing area

The FreshSatellites project, part of the DIH4AI initiative, aimed to enhance the prediction of fruit freshness and quality at the retail level using satellite imagery of the growing areas. This innovative approach sought to combine the strengths of two partners: tsenso, a startup specializing in rapid evaluation of fruit quality using their FreshIndex solution, and SDIL, experienced in working with hyperspectral satellite image data. The project planned to develop AI models to support buyers in Europe during seasonal changes in fruit sourcing regions, using advanced techniques like Bayesian neural networks and grey-box physics-supported modeling. The experiment focused on the seasonal change of mangos and red and white grapes, intending to bring more accurate and timely predictions to the food sector.