BEAF: Observing BEfore-AFter Changes to Evaluate Hallucination in Vision-language Models

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Ye-Bin, Moon, Hyeon-Woo, Nam, Choi, Wonseok, Oh, Tae-Hyun
Format: Preprint
Veröffentlicht: 2024
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866911960761630720
author Ye-Bin, Moon
Hyeon-Woo, Nam
Choi, Wonseok
Oh, Tae-Hyun
author_facet Ye-Bin, Moon
Hyeon-Woo, Nam
Choi, Wonseok
Oh, Tae-Hyun
contents Vision language models (VLMs) perceive the world through a combination of a visual encoder and a large language model (LLM). The visual encoder, pre-trained on large-scale vision-text datasets, provides zero-shot generalization to visual data, and the LLM endows its high reasoning ability to VLMs. It leads VLMs to achieve high performance on wide benchmarks without fine-tuning, exhibiting zero or few-shot capability. However, recent studies show that VLMs are vulnerable to hallucination. This undesirable behavior degrades reliability and credibility, thereby making users unable to fully trust the output from VLMs. To enhance trustworthiness and better tackle the hallucination of VLMs, we curate a new evaluation dataset, called the BEfore-AFter hallucination dataset (BEAF), and introduce new metrics: True Understanding (TU), IGnorance (IG), StuBbornness (SB), and InDecision (ID). Unlike prior works that focus only on constructing questions and answers, the key idea of our benchmark is to manipulate visual scene information by image editing models and to design the metrics based on scene changes. This allows us to clearly assess whether VLMs correctly understand a given scene by observing the ability to perceive changes. We also visualize image-wise object relationship by virtue of our two-axis view: vision and text. Upon evaluating VLMs with our dataset, we observed that our metrics reveal different aspects of VLM hallucination that have not been reported before. Project page: \url{https://beafbench.github.io/}
format Preprint
id arxiv_https___arxiv_org_abs_2407_13442
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle BEAF: Observing BEfore-AFter Changes to Evaluate Hallucination in Vision-language Models
Ye-Bin, Moon
Hyeon-Woo, Nam
Choi, Wonseok
Oh, Tae-Hyun
Computer Vision and Pattern Recognition
Computation and Language
Vision language models (VLMs) perceive the world through a combination of a visual encoder and a large language model (LLM). The visual encoder, pre-trained on large-scale vision-text datasets, provides zero-shot generalization to visual data, and the LLM endows its high reasoning ability to VLMs. It leads VLMs to achieve high performance on wide benchmarks without fine-tuning, exhibiting zero or few-shot capability. However, recent studies show that VLMs are vulnerable to hallucination. This undesirable behavior degrades reliability and credibility, thereby making users unable to fully trust the output from VLMs. To enhance trustworthiness and better tackle the hallucination of VLMs, we curate a new evaluation dataset, called the BEfore-AFter hallucination dataset (BEAF), and introduce new metrics: True Understanding (TU), IGnorance (IG), StuBbornness (SB), and InDecision (ID). Unlike prior works that focus only on constructing questions and answers, the key idea of our benchmark is to manipulate visual scene information by image editing models and to design the metrics based on scene changes. This allows us to clearly assess whether VLMs correctly understand a given scene by observing the ability to perceive changes. We also visualize image-wise object relationship by virtue of our two-axis view: vision and text. Upon evaluating VLMs with our dataset, we observed that our metrics reveal different aspects of VLM hallucination that have not been reported before. Project page: \url{https://beafbench.github.io/}
title BEAF: Observing BEfore-AFter Changes to Evaluate Hallucination in Vision-language Models
topic Computer Vision and Pattern Recognition
Computation and Language
url https://arxiv.org/abs/2407.13442