This service provides functionalities for users to train models in order to segment several types of areas related to the agriculture topic: crop types (corn, sunflower, wheat, etc), land cover (urban vs. natural, water vs land), road extraction (road vs other). The service is available as a dockerized application that can be executed in any environment with a properly configured Docker client. When starting the Docker container, user can chose between three options: preprocess the data, train a model and test a trained model.