Disentangling Knowledge-based and Visual Reasoning by Question Decomposition in KB-VQA

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Barezi, Elham J., Kordjamshidi, Parisa
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911935012798464
author Barezi, Elham J.
Kordjamshidi, Parisa
author_facet Barezi, Elham J.
Kordjamshidi, Parisa
contents We study the Knowledge-Based visual question-answering problem, for which given a question, the models need to ground it into the visual modality to find the answer. Although many recent works use question-dependent captioners to verbalize the given image and use Large Language Models to solve the VQA problem, the research results show they are not reasonably performing for multi-hop questions. Our study shows that replacing a complex question with several simpler questions helps to extract more relevant information from the image and provide a stronger comprehension of it. Moreover, we analyze the decomposed questions to find out the modality of the information that is required to answer them and use a captioner for the visual questions and LLMs as a general knowledge source for the non-visual KB-based questions. Our results demonstrate the positive impact of using simple questions before retrieving visual or non-visual information. We have provided results and analysis on three well-known VQA datasets including OKVQA, A-OKVQA, and KRVQA, and achieved up to 2% improvement in accuracy.
format Preprint
id arxiv_https___arxiv_org_abs_2406_18839
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Disentangling Knowledge-based and Visual Reasoning by Question Decomposition in KB-VQA
Barezi, Elham J.
Kordjamshidi, Parisa
Artificial Intelligence
We study the Knowledge-Based visual question-answering problem, for which given a question, the models need to ground it into the visual modality to find the answer. Although many recent works use question-dependent captioners to verbalize the given image and use Large Language Models to solve the VQA problem, the research results show they are not reasonably performing for multi-hop questions. Our study shows that replacing a complex question with several simpler questions helps to extract more relevant information from the image and provide a stronger comprehension of it. Moreover, we analyze the decomposed questions to find out the modality of the information that is required to answer them and use a captioner for the visual questions and LLMs as a general knowledge source for the non-visual KB-based questions. Our results demonstrate the positive impact of using simple questions before retrieving visual or non-visual information. We have provided results and analysis on three well-known VQA datasets including OKVQA, A-OKVQA, and KRVQA, and achieved up to 2% improvement in accuracy.
title Disentangling Knowledge-based and Visual Reasoning by Question Decomposition in KB-VQA
topic Artificial Intelligence
url https://arxiv.org/abs/2406.18839