ARTIFICIAL INTELLIGENCE PLATFORM OF GEOCUBA IN FUNCTION OF PRECISION AGRICULTURE

Abstract

In this paper are shared the main experiences of the GEOCUBA Enterprises Group in the digital transformation of the geospatial information management process of land covers based on innovation with Artificial Intelligence (AI). It is about changing the business model to deliver a systematic and automated semantic classification service for images captured from airplanes, unmanned aerial vehicles (UAVs) and earth observation satellites, such as Sentinel. To this end, a set of computer services based on technological standards has been implemented for the management of large volumes of images, the orderly management of sample taking, the training of models and their publication, the consumption of models through classification services, regression and object detection based on neural networks, and finally the delivery of the results to users through web services in a Spatial Data Infrastructure (SDI). The technological flow implemented, the new roles of the different specialists are explained and the results and difficulties faced in precision agriculture applications are exemplified, for the monitoring of the planting of various crops in the productive poles, the monitoring of the stages of the rice production campaigns and the experiments that take place in the sugarcane plantations. It is possible to provide monthly information on the planted and unplanted agricultural areas, the automated counting of the number of panicles and the humidity of the rice grain in the field, as an indicator for the harvest; population problems in sugarcane fields and the estimation of crop maturation.