MathSticks: A Benchmark for Visual Symbolic Compositional Reasoning with Matchstick Puzzles

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Ji, Yuheng, Tan, Huajie, Chi, Cheng, Xu, Yijie, Zhao, Yuting, Zhou, Enshen, Lyu, Huaihai, Wang, Pengwei, Wang, Zhongyuan, Zhang, Shanghang, Zheng, Xiaolong
Formato: Preprint
Publicado: 2025
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866909818214678528
author Ji, Yuheng
Tan, Huajie
Chi, Cheng
Xu, Yijie
Zhao, Yuting
Zhou, Enshen
Lyu, Huaihai
Wang, Pengwei
Wang, Zhongyuan
Zhang, Shanghang
Zheng, Xiaolong
author_facet Ji, Yuheng
Tan, Huajie
Chi, Cheng
Xu, Yijie
Zhao, Yuting
Zhou, Enshen
Lyu, Huaihai
Wang, Pengwei
Wang, Zhongyuan
Zhang, Shanghang
Zheng, Xiaolong
contents We introduce \textsc{MathSticks}, a benchmark for Visual Symbolic Compositional Reasoning (VSCR), which unifies visual perception, symbolic manipulation, and arithmetic consistency. Each task presents an incorrect matchstick equation that must be corrected by moving one or two sticks under strict conservation rules. The benchmark includes both text-guided and purely visual settings, systematically covering digit scale, move complexity, solution multiplicity, and operator variation, with 1.4M generated instances and a curated test set. Evaluations of 14 vision--language models reveal substantial limitations: closed-source models succeed only on simple cases, open-source models fail in the visual regime, while humans exceed 90\% accuracy. These findings establish \textsc{MathSticks} as a rigorous testbed for advancing compositional reasoning across vision and symbols. Our code and dataset are publicly available at https://github.com/Yuheng2000/MathSticks.
format Preprint
id arxiv_https___arxiv_org_abs_2510_00483
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle MathSticks: A Benchmark for Visual Symbolic Compositional Reasoning with Matchstick Puzzles
Ji, Yuheng
Tan, Huajie
Chi, Cheng
Xu, Yijie
Zhao, Yuting
Zhou, Enshen
Lyu, Huaihai
Wang, Pengwei
Wang, Zhongyuan
Zhang, Shanghang
Zheng, Xiaolong
Computer Vision and Pattern Recognition
We introduce \textsc{MathSticks}, a benchmark for Visual Symbolic Compositional Reasoning (VSCR), which unifies visual perception, symbolic manipulation, and arithmetic consistency. Each task presents an incorrect matchstick equation that must be corrected by moving one or two sticks under strict conservation rules. The benchmark includes both text-guided and purely visual settings, systematically covering digit scale, move complexity, solution multiplicity, and operator variation, with 1.4M generated instances and a curated test set. Evaluations of 14 vision--language models reveal substantial limitations: closed-source models succeed only on simple cases, open-source models fail in the visual regime, while humans exceed 90\% accuracy. These findings establish \textsc{MathSticks} as a rigorous testbed for advancing compositional reasoning across vision and symbols. Our code and dataset are publicly available at https://github.com/Yuheng2000/MathSticks.
title MathSticks: A Benchmark for Visual Symbolic Compositional Reasoning with Matchstick Puzzles
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2510.00483