In this project 4 distinct tasks (gender detection (A1), smile detection (A2), face-shape recognition (B1), eye-color recognition (B2)) are adressed following 4 different approaches and exploiting the potentialities of CNNs and HOG descriptors along with SVMs.