Answering knowledge-based visual questions via the exploration of Question Purpose

作者:

Highlights:

• Explore the purposes of visual questions and use the purposes for answer prediction.

• Estimate the correctness of each candidate answer by the consistency between candidate answers and the Question Purpose.

• Exploit the knowledge facts accordant with the Question Purpose to improve predictive results.

摘要

•Explore the purposes of visual questions and use the purposes for answer prediction.•Estimate the correctness of each candidate answer by the consistency between candidate answers and the Question Purpose.•Exploit the knowledge facts accordant with the Question Purpose to improve predictive results.

论文关键词:Visual question answering,DNN,Question Purpose

论文评审过程:Received 5 November 2021, Revised 13 July 2022, Accepted 28 August 2022, Available online 29 August 2022, Version of Record 5 September 2022.

论文官网地址:https://doi.org/10.1016/j.patcog.2022.109015