PulseAugur
EN
LIVE 08:21:59

New COMEX benchmark advances explainable AI for image cropping

Researchers have introduced COMEX, a new benchmark and learning framework designed to improve explainable aesthetic image cropping. This framework addresses the limitations of existing methods by treating explanation as an integral part of the cropping process, rather than a post-hoc addition. COMEX enables joint learning of crop localization, composition understanding, and explanation generation through a structured crop-composition-explanation problem. The proposed SFT+GRPO framework, tested on 15 large vision-language models and other cropping methods, demonstrates effectiveness and transferability across various benchmarks. AI

IMPACT Enhances AI's ability to understand and explain aesthetic choices in image manipulation, potentially improving creative tools.

RANK_REASON The cluster contains a research paper detailing a new benchmark and framework for AI image cropping. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →

New COMEX benchmark advances explainable AI for image cropping

COVERAGE [1]

  1. arXiv cs.AI TIER_1 English(EN) · Rui Yang, Wei Zhou, Dingyong Gou, Xiaohui Cui, Cong Li, Yinyin Gong, Yipo Huang, Jiliang Zhao ·

    COMEX: A Composition-Grounded Benchmark and Learning Framework for Explainable Aesthetic Image Cropping

    arXiv:2608.07570v1 Announce Type: cross Abstract: Explainable aesthetic image cropping requires not only localizing a visually pleasing crop but also explaining why it is preferred. Existing crop-and-explain methods largely treat explanation as post-hoc text generation and overlo…