Work

SciFlow-Bench: Evaluating Structure-Aware Scientific Diagram Generation via Inverse Parsing

Year: 2026

Type: conference-paper

Abstract: Scientific diagrams convey explicit structural information, yet modern text-to-image models often produce visually plausible but structurally incorrect results.Existing benchmarks either rely on image-centric or subjective metrics insensitive to structure, or evaluate intermediate symbolic representations rather than final rendered images, leaving pixel-based diagram generation underexplored.We introduce SciFlow-Bench, a structure-first benchmark for evaluating scientific diagram generation directly from pixel-level outputs.Built from real scientific PDFs, SciFlow-Bench pairs each source framework figure with a canonical ground-truth graph and evaluates models as black-box image generators under a closed-loop, round-trip protocol that inverse-parses generated diagram images back into structured graphs for comparison.This design enforces evaluation by structural recoverability rather than visual similarity alone, and is enabled by a hierarchical multiagent system that coordinates planning, perception, and structural reasoning.Experiments show that preserving structural correctness remains a fundamental challenge, particularly for diagrams with complex topology, underscoring the need for structure-aware evaluation.

Source: Proceedings of the conference - Association for Computational Linguistics. Meeting

DOI: 10.18653/v1/2026.acl-long.807

Authors: Tong Zhang, Honglin Lin, Zhou Liu, Chong Chen, Wentao Zhang

Institutions: Peking University, Shanghai Jiao Tong University, Huawei Technologies (China)

Language: English

Topic: Generative Adversarial Networks and Image Synthesis

Subfield: Computer Vision and Pattern Recognition

Field: Computer Science

Domain: Physical Sciences

Open Access status: hybrid

Funders: National Natural Science Foundation of China, Ministry of Education of the People's Republic of China, Peking University

Awards: G2981796383, G4572520933