Hallu-PI: Evaluating Hallucination in Multi-modal Large Language Models within Perturbed Inputs

Peng Ding, Jingyu Wu, Jun Kuang, Dan Ma, Xuezhi Cao, Xunliang Cai, Shi Chen, Jiajun Chen, Shujian Huang

Abstract: Multi-modal Large Language Models (MLLMs) have demonstrated remarkable performance on various visual-language understanding and generation tasks. However, MLLMs occasionally generate content inconsistent with the given images, which is known as "hallucination". Prior works primarily center on evaluating hallucination using standard, unperturbed benchmarks, which overlook the prevalent occurrence of perturbed inputs in real-world scenarios-such as image cropping or blurring-that are critical for a comprehensive assessment of MLLMs' hallucination. In this paper, to bridge this gap, we propose Hallu-PI, the first benchmark designed to evaluate Hallucination in MLLMs within Perturbed Inputs. Specifically, Hallu-PI consists of seven perturbed scenarios, containing 1,260 perturbed images from 11 object types. Each image is accompanied by detailed annotations, which include fine-grained hallucination types, such as existence, attribute, and relation. We equip these annotations with a rich set of questions, making Hallu-PI suitable for both discriminative and generative tasks. Extensive experiments on 12 mainstream MLLMs, such as GPT-4V and Gemini-Pro Vision, demonstrate that these models exhibit significant hallucinations on Hallu-PI, which is not observed in unperturbed scenarios. Furthermore, our research reveals a severe bias in MLLMs' ability to handle different types of hallucinations. We also design two baselines specifically for perturbed scenarios, namely Perturbed-Reminder and Perturbed-ICL. We hope that our study will bring researchers' attention to the limitations of MLLMs when dealing with perturbed inputs, and spur further investigations to address this issue. Our code and datasets are publicly available at https://github.com/NJUNLP/Hallu-PI.

Primary Subject Area: [Generation] Social Aspects of Generative AI

Secondary Subject Area: [Content] Vision and Language

Relevance To Conference: Our work contributes to multimedia/multimodal processing by introducing Hallu-PI, the first benchmark designed to evaluate Multi-modal Large Language Models (MLLMs) under perturbed input conditions. It tries to address the critical issue of hallucination in MLLMs, where models generate content inconsistent with the given images, which is particularly problematic in practical applications like autonomous driving systems. The benchmark includes various perturbation scenarios such as—image concatenation, image cropping, prompt misleading (and so on)-to simulate real-world conditions where inputs might be altered. Additionally, our work also proposes two baselines to mitigate the hallucinations of MLLMs for perturbed inputs. Extensive experiments conducted reveal that mainstream MLLMs exhibits significant hallucinations on Hallu-PI, providing valuable insights for researchers and developers to improve MLLMs' reliability and security.

Submission Number: 423