AI RESEARCH
CREval: An Automated Interpretable Evaluation for Creative Image Manipulation under Complex Instructions
arXiv CS.CV
•
ArXi:2603.26174v1 Announce Type: new Instruction-based multimodal image manipulation has recently made rapid progress. However, existing evaluation methods lack a systematic and human-aligned framework for assessing model performance on complex and creative editing tasks. To address this gap, we propose CREval, a fully automated question-answer (QA)-based evaluation pipeline that overcomes the incompleteness and poor interpretability of opaque Multimodal Large Language Models (MLLMs) scoring. Simultaneously, we