CVFeb 9

GeoFocus: Blending Efficient Global-to-Local Perception for Multimodal Geometry Problem-Solving

arXiv:2602.08524v1h-index: 9Has Code
Originality Incremental advance
AI Analysis

This addresses geometry challenges in AI education tools, but it is incremental as it builds on existing multimodal frameworks.

The paper tackles geometry problem-solving for Large Multimodal Models by proposing GeoFocus, which improves accuracy by 4.7% over leading specialized models through enhanced local feature coverage and reduced training time.

Geometry problem-solving remains a significant challenge for Large Multimodal Models (LMMs), requiring not only global shape recognition but also attention to intricate local relationships related to geometric theory. To address this, we propose GeoFocus, a novel framework comprising two core modules. 1) Critical Local Perceptor, which automatically identifies and emphasizes critical local structure (e.g., angles, parallel lines, comparative distances) through thirteen theory-based perception templates, boosting critical local feature coverage by 61% compared to previous methods. 2) VertexLang, a compact topology formal language, encodes global figures through vertex coordinates and connectivity relations. By replacing bulky code-based encodings, VertexLang reduces global perception training time by 20% while improving topology recognition accuracy. When evaluated in Geo3K, GeoQA, and FormalGeo7K, GeoFocus achieves a 4.7% accuracy improvement over leading specialized models and demonstrates superior robustness in MATHVERSE under diverse visual conditions. Project Page -- https://github.com/dle666/GeoFocus

Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes