mlsanityData to 14 Oct 2026

Bridging Vision, Language, and Mathematics: Pictographic Character Reconstruction with B'ezier Curves

arXiv 2511.00076 ↗First posted 29 Oct 2025cs.LGcited by 0

On the map: Multimodal LLMs / vision-language models → VLM benchmarks and visual reasoning

Abstract (arXiv)

While Vision-language Models (VLMs) have demonstrated strong semantic capabilities, their ability to interpret the underlying geometric structure of visual information is less explored. Pictographic characters, which combine visual form with symbolic structure, provide an ideal test case for this capability. We formulate this visual recognition challenge in the mathematical domain, where each character is represented by an executable program of geometric primitives. This is framed as a program synthesis task, training a VLM to decompile raster images into programs composed of B'ezier curves. Our model, acting as a "visual decompiler", demonstrates performance superior to strong zero-shot baselines, including GPT-4o. The most significant finding is that when trained solely on modern Chinese characters, the model is able to reconstruct ancient Oracle Bone Script in a zero-shot context. This generalization provides strong evidence that the model acquires an abstract and transferable geometric grammar, moving beyond pixel-level pattern recognition to a more structured form of visual understanding.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in VLM benchmarks and visual reasoning, one of 16 directions in Multimodal LLMs / vision-language models. Builds on 23 of its 34 references that are in the map. First posted on arXiv 29 Oct 2025.

Builds on

23 of 34 references.

Closest ancestors · 1

Same direction, newest first

  1. 2024Can Visual Language Models Replace OCR-Based Visual Question Answering Pipelines in Production? A Case Study in Retail28 Aug

Other references · 22

  1. 2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,744
  2. 2022Training language models to follow instructions with human feedbackcited by 15,971
  3. 2015Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networkscited by 9,639
  4. 2024DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Modelscited by 9,379
  5. 2023Judging LLM-as-a-Judge with MT-Bench and Chatbot Arenacited by 8,626
  6. 2013Rich feature hierarchies for accurate object detection and semantic segmentationcited by 2,162
  7. 2022Winoground: Probing Vision and Language Models for Visio-Linguistic Compositionalitycited by 529
  8. 2024SWIFT:A Scalable lightWeight Infrastructure for Fine-Tuningcited by 355
  9. 2020ABCNet: Real-time Scene Text Spotting with Adaptive Bezier-Curve Networkcited by 170
  10. 2017Learning to Infer Graphics Programs from Hand-Drawn Imagescited by 74
  11. 2020B'ezierSketch: A generative model for scalable vector sketchescited by 43
  12. 2024Chat2SVG: Vector Graphics Generation with Large Language Models and Image Diffusion Modelscited by 38
10 more
  1. 2024DocVLM: Make Your VLM an Efficient Readercited by 28
  2. 2024StrokeNUWA: Tokenizing Strokes for Vector Graphic Synthesiscited by 26
  3. 2025Ocean-OCR: Towards General OCR Application via a Vision-Language Modelcited by 25
  4. 2023VecFusion: Vector Font Generation with Diffusioncited by 20
  5. 2024Self-Training Large Language Models for Improved Visual Program Synthesis With Visual Reinforcementcited by 12
  6. 2024Diffusion On Syntax Trees For Program Synthesiscited by 10
  7. 2024The Impact of Visual Information in Chinese Characters: Evaluating Large Models' Ability to Recognize and Utilize Radicalscited by 10
  8. 2025OracleFusion: Assisting the Decipherment of Oracle Bone Script with Structurally Constrained Semantic Typographycited by 8
  9. 2021BezierSeg: Parametric Shape Representation for Fast Object Segmentation in Medical Imagescited by 4
  10. 2023ImageEye: Batch Image Processing Using Program Synthesiscited by 4

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2025PictOBI-20k: Unveiling Large Multimodal Models in Visual Decipherment for Pictographic Oracle Bone Characters
  2. 2026Reading or Guessing? Visual Grounding Failures of Vision-Language Models for OCR in Ancient Greek Editions
  3. 2024OracleSage: Towards Unified Visual-Linguistic Understanding of Oracle Bone Scripts through Cross-Modal Knowledge Fusion
  4. 2025Interpretable Oracle Bone Script Decipherment through Radical and Pictographic Analysis with LVLMs
  5. 2026Specializing Large Models for Oracle Bone Script Interpretation via Component-Grounded Multimodal Knowledge Augmentation
  6. 2026Enabling Stroke-Level Structural Analysis of Hieroglyphic Scripts without Language-Specific Priors
  7. 2025CalliReader: Contextualizing Chinese Calligraphy via an Embedding-Aligned Vision-Language Model
  8. 2026BareBones: Benchmarking Zero-Shot Geometric Comprehension in VLMs