A Survey on Hallucination in Large Vision-Language Models

TL;DR

This survey reviews hallucinations in large vision-language models, analyzing their causes, evaluation benchmarks, mitigation methods, and future research directions to address the misalignment between visual content and textual output.

Contribution

It provides a comprehensive overview of hallucination phenomena in LVLMs, clarifies their types, evaluates existing benchmarks and mitigation strategies, and discusses future research challenges.

Findings

Hallucinations stem from training data and model architecture issues.

Existing benchmarks for hallucination evaluation are outlined.

Current mitigation methods have limitations and need further development.

Abstract

Recent development of Large Vision-Language Models (LVLMs) has attracted growing attention within the AI landscape for its practical implementation potential. However, ``hallucination'', or more specifically, the misalignment between factual visual content and corresponding textual generation, poses a significant challenge of utilizing LVLMs. In this comprehensive survey, we dissect LVLM-related hallucinations in an attempt to establish an overview and facilitate future mitigation. Our scrutiny starts with a clarification of the concept of hallucinations in LVLMs, presenting a variety of hallucination symptoms and highlighting the unique challenges inherent in LVLM hallucinations. Subsequently, we outline the benchmarks and methodologies tailored specifically for evaluating hallucinations unique to LVLMs. Additionally, we delve into an investigation of the root causes of these hallucinations, encompassing insights from the training data and model components. We also critically review existing methods for mitigating hallucinations. The open questions and future directions pertaining to hallucinations within LVLMs are discussed to conclude this survey.