AIM 2025高FPS运动去模糊挑战赛：方法与结果
计算机视觉与模式识别
2025-09-09 v1
摘要
本文对AIM 2025高FPS非均匀运动去模糊挑战赛进行了全面回顾，重点介绍了所提出的解决方案和最终结果。该挑战赛的目标是识别有效的网络，通过学习针对复杂运动类型聚合的代表性视觉线索，在多样且具挑战性的条件下生成更清晰、视觉上更具说服力的图像。共有68名参赛者注册了比赛，最终有9个团队提交了有效参赛作品。本文全面评估了高FPS单图像运动去模糊领域的最新进展，展示了该领域的重大进步，同时利用了新数据集MIORe的样本，该数据集引入了具有挑战性的运动模式示例。
关键词
引用
@article{arxiv.2509.06793,
title = {AIM 2025 Challenge on High FPS Motion Deblurring: Methods and Results},
author = {George Ciubotariu and Florin-Alexandru Vasluianu and Zhuyun Zhou and Nancy Mehta and Radu Timofte and Ke Wu and Long Sun and Lingshun Kong and Zhongbao Yang and Jinshan Pan and Jiangxin Dong and Jinhui Tang and Hao Chen and Yinghui Fang and Dafeng Zhang and Yongqi Song and Jiangbo Guo and Shuhua Jin and Zeyu Xiao and Rui Zhao and Zhuoyuan Li and Cong Zhang and Yufeng Peng and Xin Lu and Zhijing Sun and Chengjie Ge and Zihao Li and Zishun Liao and Ziang Zhou and Qiyu Kang and Xueyang Fu and Zheng-Jun Zha and Yuqian Zhang and Shuai Liu and Jie Liu and Zhuhao Zhang and Lishen Qu and Zhihao Liu and Shihao Zhou and Yaqi Luo and Juncheng Zhou and Jufeng Yang and Qianfeng Yang and Qiyuan Guan and Xiang Chen and Guiyue Jin and Jiyu Jin},
journal= {arXiv preprint arXiv:2509.06793},
year = {2025}
}
备注
ICCVW AIM 2025