The higher availability of satellite imagery has improved the collection of information on crop conditions. This kind of data, however, only provides a valid alternative to traditional crop scouting when the image resolution is sufficiently high.
So far, unmanned aerial vehicles (UAVs) have provided better image quality than satellites and have thus been widely adopted despite the onerous data acquisition sessions necessary in the field. To reduce the resolution gap between UAV and satellite images, GENUINE adopts an innovative approach based on deep learning (DL) and the joint use of Copernicus Sentinel-2 (S2) and Planet imagery. DL is used to perform automatic boundary detection and increase the resolution of S2 (10 m) to that of Planet (≤3 m). Then the appropriate vegetation indices are computed inside the field of interest at pixel level to generate prescription maps for estimating crop vitality and the fertiliser and water required. The GENUINE solution includes web and mobile apps that enable users to select areas of interest, estimate spectral indices, and take advantage of an innovative GNSS solution for operational tracking.

Sapienza University of Rome
Lorenzo Lastilla, Matteo Amendola, Valeria Belloni, Virginia Coletta, Marco Fortunato, Valeria Marsocci, Michela Ravanelli, Roberta Ravanelli
info@genuineagriculture.com