The present work will investigate an Artificial Intelligence (AI) approach in the framework of image segmentation and clustering algorithms for star sensors applications. It focuses on the architecture development and test of a Convolutional Neural Network (CNN) based algorithm and results comparison with the state of the art. The problem of image segmentation will be faced using the U-Net to detect the brightest objects in the sensor’s Field Of View (FOV) for attitude determination purposes. The dataset creation for the network training, algorithm design process and definition of performance indices are provided together with comparison test results