# PGDP5K: A Diagram Parsing Dataset for Plane Geometry Problems

arXiv: https://arxiv.org/abs/2205.09947 · HTML: https://mlsanity.com/paper/2205.09947/

## Key facts

- Cited by 14 papers in the map (data through 14 Oct 2026).
- Builds on 5 of its 22 references that are in the map.
- First posted on arXiv 20 May 2022.

## Builds on (most cited first, 5 of 5)

- [Scene Graph Generation by Iterative Message Passing](https://mlsanity.com/paper/1701.02426/) (2017)
- [Inter-GPS: Interpretable Geometry Problem Solving with Formal Language and Symbolic Reasoning](https://mlsanity.com/paper/2105.04165/) (2021)
- [GeoQA: A Geometric Question Answering Benchmark Towards Multimodal Numerical Reasoning](https://mlsanity.com/paper/2105.14517/) (2021)
- [Fully Convolutional Scene Graph Generation](https://mlsanity.com/paper/2103.16083/) (2021)
- [Automated Generation of Geometric Theorems from Images of Diagrams](https://mlsanity.com/paper/1406.1638/) (2014)

## Built on by (most cited first, 14 of 14)

- [A Survey of Deep Learning for Mathematical Reasoning](https://mlsanity.com/paper/2212.10535/) (2022)
- [A Multi-Modal Neural Geometric Solver with Textual Clauses Parsed from Diagram](https://mlsanity.com/paper/2302.11097/) (2023)
- [MM-MATH: Advancing Multimodal Math Evaluation with Process Evaluation and Fine-grained Classification](https://mlsanity.com/paper/2404.05091/) (2024)
- [FormalGeo: An Extensible Formalized Framework for Olympiad Geometric Problem Solving](https://mlsanity.com/paper/2310.18021/) (2023)
- [Mathematical Language Models: A Survey](https://mlsanity.com/paper/2312.07622/) (2023)
- [GeoX: Geometric Problem Solving Through Unified Formalized Vision-Language Pre-training](https://mlsanity.com/paper/2412.11863/) (2024)
- [Enhancing the Geometric Problem-Solving Ability of Multimodal LLMs via Symbolic-Neural Integration](https://mlsanity.com/paper/2504.12773/) (2025)
- [TrustGeoGen: Formal-Verified Data Engine for Trustworthy Multi-modal Geometric Problem Solving](https://mlsanity.com/paper/2504.15780/) (2025)
- [Towards Geometry Problem Solving in the Large Model Era: A Survey](https://mlsanity.com/paper/2506.02690/) (2025)
- [A Survey of Deep Learning for Geometry Problem Solving](https://mlsanity.com/paper/2507.11936/) (2025)
- [FGeo-TP: A Language Model-Enhanced Solver for Geometry Problems](https://mlsanity.com/paper/2402.09047/) (2024)
- [A Survey of Multimodal Mathematical Reasoning: From Perception, Alignment to Reasoning](https://mlsanity.com/paper/2603.08291/) (2026)
- [Verifiable Geometry Problem Solving: Solver-Driven Autoformalization and Theorem Proposing](https://mlsanity.com/paper/2606.27926/) (2026)
- [Euclid-Omni : A Unified Neuro-Symbolic Framework for Plane Geometry](https://mlsanity.com/paper/2608.14585/) (2026)

## Abstract (arXiv)

Diagram parsing is an important foundation for geometry problem solving, attracting increasing attention in the field of intelligent education and document image understanding. Due to the complex layout and between-primitive relationship, plane geometry diagram parsing (PGDP) is still a challenging task deserving further research and exploration. An appropriate dataset is critical for the research of PGDP. Although some datasets with rough annotations have been proposed to solve geometric problems, they are either small in scale or not publicly available. The rough annotations also make them not very useful. Thus, we propose a new large-scale geometry diagram dataset named PGDP5K and a novel annotation method. Our dataset consists of 5000 diagram samples composed of 16 shapes, covering 5 positional relations, 22 symbol types and 6 text types. Different from previous datasets, our PGDP5K dataset is labeled with more fine-grained annotations at primitive level, including primitive classes, locations and relationships. What is more, combined with above annotations and geometric prior knowledge, it can generate intelligible geometric propositions automatically and uniquely. We performed experiments on PGDP5K and IMP-Geometry3K datasets reveal that the state-of-the-art (SOTA) method achieves only 66.07% F1 value. This shows that PGDP5K presents a challenge for future research. Our dataset is available at http://www.nlpr.ia.ac.cn/databases/CASIA-PGDP5K/.