Saved in:
Bibliographic Details
Main Authors: Sun, Jiayin, Sun, Caixia, Yang, Boyu, Li, Hailin, Chen, Xiao, Zhang, Yi, Ding, Errui, Li, Liang, Deng, Chao, Feng, Junlan
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2603.22687
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914426318225408
author Sun, Jiayin
Sun, Caixia
Yang, Boyu
Li, Hailin
Chen, Xiao
Zhang, Yi
Ding, Errui
Li, Liang
Deng, Chao
Feng, Junlan
author_facet Sun, Jiayin
Sun, Caixia
Yang, Boyu
Li, Hailin
Chen, Xiao
Zhang, Yi
Ding, Errui
Li, Liang
Deng, Chao
Feng, Junlan
contents Multimodal Large Language Models (MLLMs) have recently demonstrated remarkable perceptual and reasoning abilities. However, they struggle to perceive fine-grained geometric structures, constraining their ability of geometric understanding and visual reasoning. To address this, we propose GeoTikzBridge, a framework that enhances local geometric perception and visual reasoning through tikz-based code generation. Within this framework, we build two models supported by two complementary datasets. The GeoTikzBridge-Base model is trained on GeoTikz-Base dataset, the largest image-to-tikz dataset to date with 2.5M pairs (16 $\times$ larger than existing open-sourced datasets). This process is achieved via iterative data expansion and a localized geometric transformation strategy. Subsequently, GeoTikzBridge-Instruct is fine-tuned on GeoTikz-Instruct dataset which is the first instruction-augmented tikz dataset supporting visual reasoning. Extensive experimental results demonstrate that our models achieve state-of-the-art performance among open-sourced MLLMs. Furthermore, GeoTikzBridge models can serve as plug-and-play reasoning modules for any MLLM(LLM), enhancing reasoning performance in geometric problem-solving. Datasets and codes are publicly available at: https://github.com/sjy-1995/GeoTikzBridge.
format Preprint
id arxiv_https___arxiv_org_abs_2603_22687
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle GeoTikzBridge: Advancing Multimodal Code Generation for Geometric Perception and Reasoning
Sun, Jiayin
Sun, Caixia
Yang, Boyu
Li, Hailin
Chen, Xiao
Zhang, Yi
Ding, Errui
Li, Liang
Deng, Chao
Feng, Junlan
Computer Vision and Pattern Recognition
Multimodal Large Language Models (MLLMs) have recently demonstrated remarkable perceptual and reasoning abilities. However, they struggle to perceive fine-grained geometric structures, constraining their ability of geometric understanding and visual reasoning. To address this, we propose GeoTikzBridge, a framework that enhances local geometric perception and visual reasoning through tikz-based code generation. Within this framework, we build two models supported by two complementary datasets. The GeoTikzBridge-Base model is trained on GeoTikz-Base dataset, the largest image-to-tikz dataset to date with 2.5M pairs (16 $\times$ larger than existing open-sourced datasets). This process is achieved via iterative data expansion and a localized geometric transformation strategy. Subsequently, GeoTikzBridge-Instruct is fine-tuned on GeoTikz-Instruct dataset which is the first instruction-augmented tikz dataset supporting visual reasoning. Extensive experimental results demonstrate that our models achieve state-of-the-art performance among open-sourced MLLMs. Furthermore, GeoTikzBridge models can serve as plug-and-play reasoning modules for any MLLM(LLM), enhancing reasoning performance in geometric problem-solving. Datasets and codes are publicly available at: https://github.com/sjy-1995/GeoTikzBridge.
title GeoTikzBridge: Advancing Multimodal Code Generation for Geometric Perception and Reasoning
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2603.22687