VidHalluc: Evaluating Temporal Hallucinations in Multimodal Large Language Models for Video Understanding
- URL: http://arxiv.org/abs/2412.03735v1
- Date: Wed, 04 Dec 2024 22:03:19 GMT
- Title: VidHalluc: Evaluating Temporal Hallucinations in Multimodal Large Language Models for Video Understanding
- Authors: Chaoyu Li, Eun Woo Im, Pooyan Fazli,
- Abstract summary: We introduce VidHalluc, the largest benchmark designed to examine hallucinations in MLLMs for video understanding tasks.
VidHalluc assesses hallucinations across three critical dimensions: (1) action, (2) temporal sequence, and (3) scene transition.
We propose DINO-HEAL, a training-free method that reduces hallucinations by incorporating spatial saliency information from DINOv2 to reweight visual features during inference.
- Score: 1.1834200163382398
- License:
- Abstract: Multimodal large language models (MLLMs) have recently shown significant advancements in video understanding, excelling in content reasoning and instruction-following tasks. However, the problem of hallucination, where models generate inaccurate or misleading content, remains underexplored in the video domain. Building on the observation that the visual encoder of MLLMs often struggles to differentiate between video pairs that are visually distinct but semantically similar, we introduce VidHalluc, the largest benchmark designed to examine hallucinations in MLLMs for video understanding tasks. VidHalluc assesses hallucinations across three critical dimensions: (1) action, (2) temporal sequence, and (3) scene transition. VidHalluc consists of 5,002 videos, paired based on semantic similarity and visual differences, focusing on cases where hallucinations are most likely to occur. Through comprehensive testing, our experiments show that most MLLMs are vulnerable to hallucinations across these dimensions. Furthermore, we propose DINO-HEAL, a training-free method that reduces hallucinations by incorporating spatial saliency information from DINOv2 to reweight visual features during inference. Our results demonstrate that DINO-HEAL consistently improves performance on VidHalluc, achieving an average improvement of 3.02% in mitigating hallucinations among all tasks. Both the VidHalluc benchmark and DINO-HEAL code can be accessed via $\href{https://vid-halluc.github.io/}{\text{this link}}$.
Related papers
- VidHal: Benchmarking Temporal Hallucinations in Vision LLMs [9.392258475822915]
We introduce VidHal, a benchmark specially designed to evaluate video-based hallucinations.
A defining feature of VidHal is the careful creation of captions which represent varying levels of captions associated with each video.
We propose a novel caption ordering task requiring VLLMs to rank captions by hallucinatory extent.
arXiv Detail & Related papers (2024-11-25T06:17:23Z) - Hallu-PI: Evaluating Hallucination in Multi-modal Large Language Models within Perturbed Inputs [54.50483041708911]
Hallu-PI is the first benchmark designed to evaluate hallucination in MLLMs within Perturbed Inputs.
Hallu-PI consists of seven perturbed scenarios, containing 1,260 perturbed images from 11 object types.
Our research reveals a severe bias in MLLMs' ability to handle different types of hallucinations.
arXiv Detail & Related papers (2024-08-02T16:07:15Z) - Investigating and Mitigating the Multimodal Hallucination Snowballing in Large Vision-Language Models [33.19894606649144]
Though advanced in understanding visual information with human languages, Large Vision-Language Models (LVLMs) still suffer from multimodal hallucinations.
We propose a framework called MMHalball to evaluate LVLMs' behaviors when encountering generated hallucinations.
We propose a training-free method called Residual Visual Decoding, where we revise the output distribution of LVLMs with the one derived from the residual visual input.
arXiv Detail & Related papers (2024-06-30T03:04:11Z) - VideoHallucer: Evaluating Intrinsic and Extrinsic Hallucinations in Large Video-Language Models [59.05674402770661]
This work introduces VideoHallucer, the first comprehensive benchmark for hallucination detection in large video-language models (LVLMs)
VideoHallucer categorizes hallucinations into two main types: intrinsic and extrinsic, offering further subcategories for detailed analysis.
arXiv Detail & Related papers (2024-06-24T06:21:59Z) - The Dawn After the Dark: An Empirical Study on Factuality Hallucination
in Large Language Models [134.6697160940223]
hallucination poses great challenge to trustworthy and reliable deployment of large language models.
Three key questions should be well studied: how to detect hallucinations (detection), why do LLMs hallucinate (source), and what can be done to mitigate them.
This work presents a systematic empirical study on LLM hallucination, focused on the the three aspects of hallucination detection, source and mitigation.
arXiv Detail & Related papers (2024-01-06T12:40:45Z) - Hallucination Augmented Contrastive Learning for Multimodal Large
Language Model [53.65682783591723]
Multi-modal large language models (MLLMs) have been shown to efficiently integrate natural language with visual information to handle multi-modal tasks.
However, MLLMs still face a fundamental limitation of hallucinations, where they tend to generate erroneous or fabricated information.
In this paper, we address hallucinations in MLLMs from a novel perspective of representation learning.
arXiv Detail & Related papers (2023-12-12T04:05:15Z) - Analyzing and Mitigating Object Hallucination in Large Vision-Language Models [110.12460299261531]
Large vision-language models (LVLMs) have shown remarkable abilities in understanding visual information with human languages.
LVLMs still suffer from object hallucination, which is the problem of generating descriptions that include objects that do not actually exist in the images.
We propose a powerful algorithm, LVLM Hallucination Revisor (LURE), to rectify object hallucination in LVLMs by reconstructing less hallucinatory descriptions.
arXiv Detail & Related papers (2023-10-01T18:10:53Z) - Evaluating Object Hallucination in Large Vision-Language Models [122.40337582958453]
This work presents the first systematic study on object hallucination of large vision-language models (LVLMs)
We find that LVLMs tend to generate objects that are inconsistent with the target images in the descriptions.
We propose a polling-based query method called POPE to evaluate the object hallucination.
arXiv Detail & Related papers (2023-05-17T16:34:01Z)
This list is automatically generated from the titles and abstracts of the papers in this site.
This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.