Git Loss for Deep Face Recognition
share
Convolutional Neural Networks (CNNs) have been widely used in computer vision tasks, such as face recognition, and have achieved state-of-the-art results due to their ability to learn discriminative deep features. Conventionally, CNNs have been trained with Softmax as supervision signal to penalize the classification loss. In order to further enhance the discriminative capability of deep features, we introduced a joint supervision signal, Git loss, which leverages on Softmax and Center loss functions. The aim of our loss function is to minimizes the intra-class variances as well as maximizes the inter-class distances. Such minimization and maximization of deep features are considered ideal for face recognition task. Results obtained on two popular face recognition benchmarks datasets show that our proposed loss function achieves maximum separability between deep face features of different identities and achieves state-of-the-art accuracy on two major face recognition benchmark datasets: Labeled Faces in the Wild (LFW) and YouTube Faces (YTF).
READ FULL TEXT