CodePercept: Code-Grounded Visual STEM Perception for MLLMs

TL;DR

CodePercept introduces a novel approach to enhance STEM visual perception in multimodal large language models by leveraging executable code as a perceptual medium, supported by a large dataset and a new evaluation benchmark.

Contribution

It systematically demonstrates perception as the key bottleneck in STEM visual reasoning and proposes code-grounded perception methods with a new dataset and benchmark.

Findings

Perception scaling outperforms reasoning scaling in STEM visual tasks.

Code-grounded captioning reduces hallucinations in image descriptions.

The new benchmark enables deterministic evaluation of visual perception in STEM.

Abstract

When MLLMs fail at Science, Technology, Engineering, and Mathematics (STEM) visual reasoning, a fundamental question arises: is it due to perceptual deficiencies or reasoning limitations? Through systematic scaling analysis that independently scales perception and reasoning components, we uncover a critical insight: scaling perception consistently outperforms scaling reasoning. This reveals perception as the true lever limiting current STEM visual reasoning. Motivated by this insight, our work focuses on systematically enhancing the perception capabilities of MLLMs by establishing code as a powerful perceptual medium--executable code provides precise semantics that naturally align with the structured nature of STEM visuals. Specifically, we construct ICC-1M, a large-scale dataset comprising 1M Image-Caption-Code triplets that materializes this code-as-perception paradigm through two complementary approaches: (1) Code-Grounded Caption Generation treats executable code as ground truth for image captions, eliminating the hallucinations inherent in existing knowledge distillation methods; (2) STEM Image-to-Code Translation prompts models to generate reconstruction code, mitigating the ambiguity of natural language for perception enhancement. To validate this paradigm, we further introduce STEM2Code-Eval, a novel benchmark that directly evaluates visual perception in STEM domains. Unlike existing work relying on problem-solving accuracy as a proxy that only measures problem-relevant understanding, our benchmark requires comprehensive visual comprehension through executable code generation for image reconstruction, providing deterministic and verifiable assessment. Code is available at https://github.com/TongkunGuan/Qwen-CodePercept.