CHAOS: Chart Analysis with Outlier Samples

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Moured, Omar, Chen, Yufan, Liu, Ruiping, Reiß, Simon, Torr, Philip, Zhang, Jiaming, Stiefelhagen, Rainer
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910964102725632
author Moured, Omar
Chen, Yufan
Liu, Ruiping
Reiß, Simon
Torr, Philip
Zhang, Jiaming
Stiefelhagen, Rainer
author_facet Moured, Omar
Chen, Yufan
Liu, Ruiping
Reiß, Simon
Torr, Philip
Zhang, Jiaming
Stiefelhagen, Rainer
contents Charts play a critical role in data analysis and visualization, yet real-world applications often present charts with challenging or noisy features. However, "outlier charts" pose a substantial challenge even for Multimodal Large Language Models (MLLMs), which can struggle to interpret perturbed charts. In this work, we introduce CHAOS (CHart Analysis with Outlier Samples), a robustness benchmark to systematically evaluate MLLMs against chart perturbations. CHAOS encompasses five types of textual and ten types of visual perturbations, each presented at three levels of severity (easy, mid, hard) inspired by the study result of human evaluation. The benchmark includes 13 state-of-the-art MLLMs divided into three groups (i.e., general-, document-, and chart-specific models) according to the training scope and data. Comprehensive analysis involves two downstream tasks (ChartQA and Chart-to-Text). Extensive experiments and case studies highlight critical insights into robustness of models across chart perturbations, aiming to guide future research in chart understanding domain. Data and code are publicly available at: http://huggingface.co/datasets/omoured/CHAOS.
format Preprint
id arxiv_https___arxiv_org_abs_2505_17235
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle CHAOS: Chart Analysis with Outlier Samples
Moured, Omar
Chen, Yufan
Liu, Ruiping
Reiß, Simon
Torr, Philip
Zhang, Jiaming
Stiefelhagen, Rainer
Computer Vision and Pattern Recognition
Computation and Language
Charts play a critical role in data analysis and visualization, yet real-world applications often present charts with challenging or noisy features. However, "outlier charts" pose a substantial challenge even for Multimodal Large Language Models (MLLMs), which can struggle to interpret perturbed charts. In this work, we introduce CHAOS (CHart Analysis with Outlier Samples), a robustness benchmark to systematically evaluate MLLMs against chart perturbations. CHAOS encompasses five types of textual and ten types of visual perturbations, each presented at three levels of severity (easy, mid, hard) inspired by the study result of human evaluation. The benchmark includes 13 state-of-the-art MLLMs divided into three groups (i.e., general-, document-, and chart-specific models) according to the training scope and data. Comprehensive analysis involves two downstream tasks (ChartQA and Chart-to-Text). Extensive experiments and case studies highlight critical insights into robustness of models across chart perturbations, aiming to guide future research in chart understanding domain. Data and code are publicly available at: http://huggingface.co/datasets/omoured/CHAOS.
title CHAOS: Chart Analysis with Outlier Samples
topic Computer Vision and Pattern Recognition
Computation and Language
url https://arxiv.org/abs/2505.17235