Abstract
For highly occluded faces, only few features exist, which makes such face detection more challenging. In this paper, we propose a novel algorithm to make full use of the facial features. The proposed algorithm is based on Region-based Fully Convolutional Network (R-FCN) with two improved parts for robust face detection, including the multi-scale training and a new feature-fusion scheme. Firstly, instead of utilizing fixed scales for all faces, we adopt multi-scale inputs to strengthen the features of the partial faces and increase the training set diversity. Up-sampling the training images can efficiently enlarge the features of the occluded faces. Secondly, we make a feature fusion by combining layers with different sizes of receptive fields, which can preserve the details of the faces with only partial faces available. Our method achieves superior accuracy over the stat-of-the-art techniques on massively-benchmarked face dataset (WIDER FACE), and shows great improvements for highly occluded face detection.
Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
