基于多层次特征融合的Transformer人脸识别方法

Transformer face recognition method based on multi-level feature fusion

摘要:

卷积神经网络中的卷积操作只能捕获局部信息，而Transformer能保留更多的空间信息且能建立图像的长距离连接.在视觉领域的应用中，Transformer缺乏灵活的图像尺寸及特征尺度适应能力，通过利用层级式网络增强不同尺度建模的灵活性，且引入多尺度特征融合模块丰富特征信息.本文提出了一种基于改进的Swin Transformer人脸模型——Swin Face模型.Swin Face以Swin Transformer为骨干网络，引入多层次特征融合模块，增强了模型对人脸的特征表达能力，并使用联合损失函数优化策略设计人脸识别分类器，实现人脸识别.实验结果表明，与多种人脸识别方法相比，Swin Face模型通过使用分级特征融合网络，在LFW、CALFW、AgeDB-30、CFP数据集上均取得最优的效果，验证了此模型具有良好的泛化性和鲁棒性.

Abstract:

The convolutional operation in a convolutional neural network only captures local information, whereas the Transformer retains more spatial information and can create long-range connections of images. In the application of vision field, Transformer lacks flexible image size and feature scale adaptation capability. To solve these problems, the flexibility of modeling at different scales is enhanced by using hierarchical networks, and a multi-scale feature fusion module is introduced to enrich feature information. This paper proposes an improved Swin Face model based on the Swin Transformer model. The model uses the Swin Transformer as the backbone network and a multi-level feature fusion model is introduced to enhance the feature representation capability of the Swin Face model for human faces. A joint loss function optimisation strategy is used to design a face recognition classifier to realize face recognition. The experimental results show that, compared with various face recognition methods, the Swin Face recognition method achieves the best results on LFW, CALFW, AgeDB-30, and CFP datasets by using a hierarchical feature fusion network, and also has good generalization and robustness.