Knowledge base graph embedding module design for Visual question answering model

作者：

Highlights：

• A knowledge base graph embedding module is constructed to extend the generality of the visual question answering model based on knowledge base in this paper.

• This paper constructs two experimental knowledge bases: DBV and DBA through a series of knowledge base preprocessing methods.

• The DBV knowledge base constructed in this paper has strong entity correspondence, and its knowledge base embedding effect is very good.

摘要

•A knowledge base graph embedding module is constructed to extend the generality of the visual question answering model based on knowledge base in this paper.•This paper constructs two experimental knowledge bases: DBV and DBA through a series of knowledge base preprocessing methods.•The DBV knowledge base constructed in this paper has strong entity correspondence, and its knowledge base embedding effect is very good.

论文关键词：Faster R-CNN,DBpedia spotlight,knowledge base,VQA

论文评审过程：Received 11 August 2020, Revised 2 July 2021, Accepted 3 July 2021, Available online 13 July 2021, Version of Record 17 July 2021.

论文官网地址：https://doi.org/10.1016/j.patcog.2021.108153