arXiv cs.AI· Junseong Shin, Hyeonsu Jo, Daehyun Kim, Tae Hyun Kim·· 12 小时前AI 评分32
PickMoment：用去模糊与模糊到视频学习实现连续时间单图生视频
PickMoment: Continuous-Time Single-Image-to-Video via Learning Deblurring and Blur-to-Video
AI 导读
PickMoment 提出连续时间重构方法，用单一确定性模型直接学习曝光任意子区间上的区间均值模糊，将单图去模糊、模糊到视频生成和连续时间选帧统一为同一网络的不同查询，无需分任务训练。模型在 GoPro 和 HIDE 上取得生成式去模糊方法中的 SOTA，在 RealBlur 上媲美复原类方法，并在 GoPro-7 模糊到视频任务上取得最高逐帧保真度，全程单次前向传播、无需迭代采样。
正文 · 原文
Abstract:Motion blur arises from the temporal integration of a continuous sharp signal over a finite exposure window, yet existing learning-based methods sidestep this physical model and predict only the sharp signal itself: most single-image deblurring methods recover a single frame at the exposure center, while blur-to-video methods predict a fixed set of frames. We introduce PickMoment, a continuous-time reformulation that directly learns the interval-mean blur over arbitrary sub-intervals of the exposure with a single deterministic model. Drawing an analogy to MeanFlow's average-velocity formulation, we train the model with three supervisions derived from the blur integral: an empirical reconstruction loss from available subframes, an additivity loss that enforces self-consistency across overlapping sub-intervals, and a sharp-frame loss anchored at the zero-interval limit. A single trained model unifies single-image deblurring, blur-to-video generation, and continuous-time pick-a-moment recovery as different queries to the same network, with no separate training for each task. Our PickMoment achieves state-of-the-art performance among generative-based deblurring methods on GoPro and HIDE while competitive against restoration-based methods on RealBlur, and the highest per-frame fidelity on GoPro-7 blur-to-video, all in a single forward pass without iterative sampling.
| Subjects: | Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2610.01279 [cs.CV] |
| (or arXiv:2610.01279v2 [cs.CV] for this version) | |
| https://doi.org/10.48550/arXiv.2610.01279 arXiv-issued DOI via DataCite |
Submission history
From: Junseong Shin [view email]
[v1]
Thu, 1 Oct 2026 08:16:14 UTC (6,282 KB)
[v2]
Thu, 5 Nov 2026 14:20:10 UTC (6,299 KB)
来源：arXiv cs.AI · arxiv.org