基于身份感知模型的素描人脸识别方法

针对素描图像和光学图像之间存在较大的模态差异这一问题，提出了一种基于身份感知模型的素描人脸识别方法，实现跨模态图像生成和素描人脸识别。该方法应用新的感知损失来监督图像生成网络，生成更好的跨模态图像，减少模态差异带来的识别精度损失，并通过三元组损失来正则化类内和类间距离，增强识别模型的性能，用联合训练策略提升素描人脸识别能力。在UoM-SGFSv2、e-PRIP等素描人脸数据集上的实验结果表明，该方法识别效果优于其他对比算法。

Sketch face recognition based on identity perception model

To reduce domain gap between the sketches and the photos,a sketch face recognition algorithm based on the identity perception model is proposed to achieve cross-modal image generation and sketch face recognition.This algorithm applies a new perceptual loss to supervise the image generation network,generate better cross-modal images,and reduce the recognition accuracy loss caused by the modal difference.It regularizes the intra- and inter-class distances by a triplet loss,enhances the performance of the recognition model,and uses the joint training strategy to improve the ability of sketch face recognition.The experimental results on sketch face data sets such as UoM-SGFSv2 show that the proposed method is better than other state-of-the-art methods.