Gender classification aims at recognizing a person's gender. Despite the highaccuracy achieved by state-of-the-art methods for this task, there is stillroom for improvement in generalized and unrestricted datasets. In this paper,we advocate a new strategy inspired by the behavior of humans in genderrecognition. Instead of dealing with the face image as a sole feature, we relyon the combination of isolated facial features and a holistic feature which wecall the foggy face. Then, we use these features to train deep convolutionalneural networks followed by an AdaBoost-based score fusion to infer the finalgender class. We evaluate our method on four challenging datasets todemonstrate its efficacy in achieving better or on-par accuracy withstate-of-the-art methods. In addition, we present a new face dataset thatintensifies the challenges of occluded faces and illumination changes, which webelieve to be a much-needed resource for gender classification research.
展开▼