Human-robot interaction is at the core of social robotics. As such, making the robot aware of the users being aroud it is a key functionality. The problem is twofold: user detection and recognition go hand-by-hand. In this work, we present the development of a gender recognition functionality for the social robot Mopi. Face detection is made thanks to a Viola Jones detector. Gender recognition can alternatively use three popular techniques: Eigenfaces, Fisherfaces, or LBPH. The training set is built thanks to an innovative technique, using automatic image retrieval techniques. Once the classifier trained, the recognition of the gender of a new user does not require any previous knowledge about her or specific training phase. Extensive experimental results are given for verifying the usefulness of the proposed method.