Geometric prior guided hybrid deep neural network for facial beauty analysis

作     者：Tianhao Peng Mu Li Fangmei Chen Yong Xu David Zhang 

作者机构：The School of Computer Science and TechnologyGuizhou UniversityGuiyangChina Department of AutomationMoutai InstituteRenhuaiGuizhouChina The School of Computer Science and TechnologyHarbin Institute of TechnologyShenzhenShenzhenChina The Information and Communication Engineering DepartmentDalian Minzu UniversityDalianChina The School of Data ScienceThe Chinese University of Hong KongShenzhenShenzhenChina 

出 版 物：《CAAI Transactions on Intelligence Technology》 (智能技术学报（英文）)

年 卷 期：2024年第9卷第2期

页      面：467-480页

核心收录：

学科分类：12[管理学] 1201[管理学-管理科学与工程(可授管理学、工学学位)] 081104[工学-模式识别与智能系统] 08[工学] 0835[工学-软件工程] 0811[工学-控制科学与工程] 0812[工学-计算机科学与技术（可授工学、理学学位）] 

基　　金：Shenzhen Science and Technology Program,Grant/Award Number:ZDSYS20211021111415025 Shenzhen Institute of Artificial Intelligence and Robotics for Society Youth Science and Technology Talents Development Project of Guizhou Education Department,Grant/Award Number:QianJiaoheKYZi459 

主　　题：deep neural networks face analysis face biometrics image analysis 

摘      要：Facial beauty analysis is an important topic in human *** may be used as a guidance for face beautification applications such as cosmetic *** neural networks(DNNs)have recently been adopted for facial beauty analysis and have achieved remarkable ***,most existing DNN-based models regard facial beauty analysis as a normal classification *** ignore important prior knowledge in traditional machine learning models which illustrate the significant contribution of the geometric features in facial beauty *** be specific,landmarks of the whole face and facial organs are introduced to extract geometric features to make the *** by this,we introduce a novel dual-branch network for facial beauty analysis:one branch takes the Swin Transformer as the backbone to model the full face and global patterns,and another branch focuses on the masked facial organs with the residual network to model the local patterns of certain facial ***,the designed multi-scale feature fusion module can further facilitate our network to learn complementary semantic information between the two *** model optimisation,we propose a hybrid loss function,where especially geometric regulation is introduced by regressing the facial landmarks and it can force the extracted features to convey facial geometric *** performed on the SCUT-FBP5500 dataset and the SCUT-FBP dataset demonstrate that our model outperforms the state-of-the-art convolutional neural networks models,which proves the effectiveness of the proposed geometric regularisation and dual-branch structure with the hybrid *** the best of our knowledge,this is the first study to introduce a Vision Transformer into the facial beauty analysis task.