基于人脸识别的多模态人物信息补全系统设计

Design of Person Information Knowledge Base Completion System Based on Face Detection

随着知识库在各领域任务的广泛应用,知识库补全技术的作用日益凸显。考虑到人脸特征的独特性,设计并实现一种基于人脸识别的人物信息知识库补全系统。该系统首先采用基于MTCNN模型的人脸识别模块对输入的新闻图像文本对中的图像信息进行人脸检测并提取其人脸特征;其次利用基于BiLSTM的文本分析模块对新闻文本进行分词与命名实体识别;最后输入基于Insightface的信息对齐模块从而与基于MongoDB创建的人物信息知识库内人物实体进行对齐并实现知识库补全。实验证明该系统能有效提高知识库的完整性与丰富性。

With the wide application of knowledge base in various fields of deep learning,knowledge completion technology has gradually attracted the attention of more and more researchers.Based on the uniqueness of face features,this paper designed and implemented a person information knowledge base completion system based on face detection.Firstly,the face detection module based on MCTNN model is used to detect the image information in the input news image text pair and extract its face features.Secondly,the text analysis module based on BiLSTM is used to segment the news text and recognize the named entity,Finally,input the information alignment module based on Insightface to align and complete the character entities in the character information knowledge base created based on MongoDB.Experiments show that the method based on face features effectively improves the integrity and richness of knowledge base.