Pixels or Code? Benchmarking Instruction-Guided Visual Artifact Operation

arXiv:2610.16444 · cs.CV · Submitted 2026-10-12 · Read on arXiv

cs.CV

Submitted: 2026-10-12

Updated: 2026-10-12

Comments: Code: https://github.com/BearGry/VAO-Bench

Code: https://github.com/BearGry/VAO-Bench

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Terminology

• Benchmarking — 1× in this paper · explained in 4 episodes

Sources

• EdiTikZ: Scientific Figure Editing from Revision Trajectories
• ChartEditBench: Evaluating Grounded Multi-Turn Chart Editing in Multimodal Language Models
• GRADE: Benchmarking Discipline-Informed Reasoning in Image Editing
• SVGEditBench V2: A Benchmark for Instruction-based SVG Editing
• VisEditBench: Can Vision-Language Models Edit Visualization Code from Multimodal Feedback?
• SciDiagramEdit: Learning to Edit Scientific Diagrams from Paper Revisions
• DesignBench: A Comprehensive Benchmark for MLLM-based Front-end Code Generation
• ImgEdit: A Unified Image Editing Dataset and Benchmark
• ChartSync: A Benchmark for Visuo-Logical Cascading Chart Editing
• Edit2TikZ: A Comprehensive and Challenging Benchmark for Scientific Figure Editing with TikZ
• ChartEdit: How Far Are MLLMs From Automating Chart Analysis? Evaluating MLLMs' Capability via Chart Editing

Related papers

• Loss Knows Best: Detecting Annotation Errors in Videos via Loss Trajectories
• AnchorWeave: World-Consistent Video Generation with Retrieved Local Spatial Memories
• Benchmarking the Robustness of Foundation Models for Mammography under Domain Shift
• MambaX-Net: Dual-Input Mamba-Enhanced Cross-Attention Network for Longitudinal MRI Segmentation
• TeleOCR: Navigating Document Parsing Across Digital and Camera-Captured Documents
• A Survey on Efficient Vision-Language-Action Models