LoViF 2026 真实世界一体化图像复原挑战赛：方法与结果
计算机视觉与模式识别
2026-04-22 v1
摘要
本文对 LoViF 真实世界一体化图像复原挑战赛进行了综述。该挑战赛旨在推进多种真实世界退化条件（包括模糊、低光照、雾霾、雨和雪）下的真实世界一体化图像复原研究。它提供了一个统一的基准，以在共同框架下评估复原模型在多个退化类别上的鲁棒性和泛化能力。比赛吸引了 124 名注册参赛者，并收到了 9 份附有相应说明表的有效最终提交，为真实世界一体化图像复原的进展做出了重大贡献。本报告对所提交的方法及相应结果进行了详细分析，强调了统一真实世界图像复原的最新进展。该分析突出了有效方法，并为未来真实世界低级视觉研究建立了基准。
引用
@article{arxiv.2604.19445,
title = {LoViF 2026 Challenge on Real-World All-in-One Image Restoration: Methods and Results},
author = {Xiang Chen and Hao Li and Jiangxin Dong and Jinshan Pan and Xin Li and Xin He and Naiwei Chen and Shengyuan Li and Fengning Liu and Haoyi Lv and Haowei Peng and Yilian Zhong and Yuxiang Chen and Shibo Yin and Yushun Fang and Xilei Zhu and Yahui Wang and Chen Lu and Kaibin Chen and Xu Zhang and Xuhui Cao and Jiaqi Ma and Ziqi Wang and Shengkai Hu and Yuning Cui and Huan Zhang and Shi Chen and Bin Ren and Lefei Zhang and Guanglu Dong and Qiyao Zhao and Tianheng Zheng and Chunlei Li and Lichao Mou and Chao Ren and Wangzhi Xing and Xin Lu and Enxuan Gu and Jingxi Zhang and Diqi Chen and Qiaosi Yi and Bingcai Wei and Mingyu Liu and Pengyu Wang and Ce Liu and Miaoxin Guan and Boyu Chen and Hongyu Li and Jian Zhu and Xinrui Luo and Ziyang He and Jiayu Wang and Yichen Xiang and Huayi Qi and Haoyu Bian and Yiran Li and Sunlichen Zhou},
journal= {arXiv preprint arXiv:2604.19445},
year = {2026}
}
备注
CVPR Workshops 2026; https://lowlevelcv.com/