FlashVSR：面向实时扩散式流式视频超分辨率
FlashVSR: Towards Real-time Diffusion-Based Streaming Video Super Resolution
面向实时视频超分，提出三阶段蒸馏流水线把多步扩散压缩为单步流式推理，并用局部约束稀疏注意力弥合训练-测试分辨率差距；做扩散加速或高分辨率视频复原时可直接借鉴这两套机制。
原文依据 a locality-constrained sparse attention that reduces redundant computation while bridging the train--test resolution gap
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论扩散模型近期推动了视频复原的发展，但将其应用于真实世界与 AIGC 生成视频的超分辨率仍面临高延迟、计算代价过高以及对超高分辨率泛化能力差等挑战。本工作旨在通过实现高效性、可扩展性与接近实时的性能，使基于扩散的视频超分辨率走向实用。为此，我们提出 FlashVSR，这是首个面向高效视频超分辨率的基于扩散的单步流式框架。FlashVSR 在单块 A100 GPU 上对 768x1408 视频达到 17 FPS，其性能得益于三项互补的创新：(i) 一条对训练友好的三阶段蒸馏流水线，使流式超分辨率成为可能；(ii) 局部性约束的稀疏注意力，在减少冗余计算的同时弥合训练与测试之间的分辨率差距；(iii) 一个微型条件解码器，在不牺牲质量的前提下加速重建。为支持大规模训练，我们还构建了 VSR-120K，这是一个包含 12 万段视频与 18 万张图像的新数据集。大量实验表明，FlashVSR 能够可靠地扩展到超高分辨率，并取得最优性能，相比此前基于单步扩散的视频超分辨率模型最高实现 12 倍加速。我们将公开代码、预训练模型与数据集，以促进高效扩散式视频超分辨率的未来研究。
英文原文摘要 1402 字符
Diffusion models have recently advanced video restoration, but applying them to real-world and AIGC-generated video super-resolution (VSR) remains challenging due to high latency, prohibitive computation, and poor generalization to ultra-high resolutions. Our goal in this work is to make diffusion-based VSR practical by achieving efficiency, scalability, and near real-time performance. To this end, we propose FlashVSR, the first diffusion-based one-step streaming framework for efficient video super-resolution. FlashVSR runs at 17 FPS for 768x1408 videos on a single A100 GPU by combining three complementary innovations: (i) a train-friendly three-stage distillation pipeline that enables streaming super-resolution, (ii) locality-constrained sparse attention that reduces redundant computation while bridging the train--test resolution gap, and (iii) a tiny conditional decoder that accelerates reconstruction without sacrificing quality. To support large-scale training, we also construct VSR-120K, a new dataset containing 120 K videos and 180 K images. Extensive experiments demonstrate that FlashVSR scales reliably to ultra-high resolutions and achieves state-of-the-art performance with up to 12x speed-up over prior one-step diffusion-based VSR models. We will release the code, pretrained models, and dataset to foster future research in efficient diffusion-based video super-resolution.