A State-of-the-Art Computer Vision Adopting Non-Euclidean Deep-Learning Models