TASK	DATASET	MODEL	METRIC NAME	METRIC VALUE	GLOBAL RANK	REMOVE
Visual Question Answering (VQA)	A-OKVQA	VLC-BERT	DA VQA Score	38.05	# 8
Visual Question Answering (VQA)	OK-VQA	VLC-BERT	Accuracy	43.1	# 24

Badge	Markdown
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/vlc-bert-visual-question-answering-with/visual-question-answering-on-a-okvqa)](https://paperswithcode.com/sota/visual-question-answering-on-a-okvqa?p=vlc-bert-visual-question-answering-with)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/vlc-bert-visual-question-answering-with/visual-question-answering-on-ok-vqa)](https://paperswithcode.com/sota/visual-question-answering-on-ok-vqa?p=vlc-bert-visual-question-answering-with)`

VLC-BERT: Visual Question Answering with Contextualized Commonsense Knowledge

24 Oct 2022 · Sahithya Ravi, Aditya Chinchure, Leonid Sigal, Renjie Liao, Vered Shwartz ·

There has been a growing interest in solving Visual Question Answering (VQA) tasks that require the model to reason beyond the content present in the image. In this work, we focus on questions that require commonsense reasoning. In contrast to previous methods which inject knowledge from static knowledge bases, we investigate the incorporation of contextualized knowledge using Commonsense Transformer (COMET), an existing knowledge model trained on human-curated knowledge bases. We propose a method to generate, select, and encode external commonsense knowledge alongside visual and textual cues in a new pre-trained Vision-Language-Commonsense transformer model, VLC-BERT. Through our evaluation on the knowledge-intensive OK-VQA and A-OKVQA datasets, we show that VLC-BERT is capable of outperforming existing models that utilize static knowledge bases. Furthermore, through a detailed analysis, we explain which questions benefit, and which don't, from contextualized commonsense knowledge from COMET.

PDF Abstract