FaceHunter
Image Communication(2016)
摘要
In this paper, we propose a new multi-task Convolutional Neural Network (CNN) based face detector, which is named FaceHunter for simplicity. The main idea is to make the face detector achieve a high detection accuracy and obtain much reliable face boxes. Reliable face boxes output will be much helpful for further face image analysis. To reach this goal, we design a deep CNN network with a multi-task loss, i.e., one is for discriminating face and non-face, and another is for face box regression. An adaptive pooling layer is added before full connection to make the network adaptive to variable candidate proposals, and the truncated SVD is applied to compress the parameters of the fully connected layers. To further speed up the detector, the convolutional feature map is directly used to generate the candidate proposals by using Region Proposal Network (RPN). The proposed FaceHunter is evaluated on the AFW dataset, FDDB dataset and Pascal Faces respectively, and extensive experiments demonstrate its powerful performance against several state-of-the-art detectors. HighlightsThe multi-task CNN is applied in the face detection task, and its validated to be very efficient. An adaptive pooling layer is integrated into the network to make it more flexible, and also the truncated SVD is introduced to compress the fully-connected layers.The RPN network is introduced to generate the proposals, which is directly performed on the convolutional feature maps. It shares the same features with multi-task CNN, so the proposal generating cost is very small.The proposed FaceHunter is evaluated on the AFW, FDDB and Pascal Faces respectively, and extensive experiments demonstrate its powerful performance against several state-of-the-art detectors.
更多查看译文
关键词
Face detection,Convolutional neural network,Multi-task,Adaptive pooling layer,Region proposal network
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络