MS-Temba：用于理解长未剪辑视频的多尺度时序 Mamba
MS-Temba: Multi-Scale Temporal Mamba for Understanding Long Untrimmed Videos
把 Mamba 状态空间模型用空洞（dilated）SSM 扩展到多尺度时序建模，为长视频密集重叠动作的边界定位提供了可复用的骨干设计；做长时序视频理解或需要长程依赖+细粒度细节兼顾的任务时，可借鉴其多尺度 SSM 分块与轻量融合器结构。
原文依据 we propose Multi-Scale Temporal Mamba (MS-Temba), which extends Mamba to TAD with newly introduced dilated SSMs
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论未剪辑视频中的时序动作检测（TAD）面临显著挑战，尤其是日常生活活动（ADL）场景，要求模型能够：（1）处理长时长视频；（2）捕捉动作的时序变化；（3）同时检测密集重叠的动作。现有的基于 CNN 与 Transformer 的方法难以在大规模下同时捕捉细粒度细节与长程结构。基于状态空间模型（SSM）的 Mamba 具备强大的长程建模能力，但将其朴素地应用于 TAD 会破坏细粒度时序结构，且未能考虑 TAD 固有的挑战。为此，我们提出多尺度时序 Mamba（MS-Temba），通过新引入的空洞 SSM 将 Mamba 扩展到 TAD。每个 Temba 块由空洞 SSM 与我们提出的额外损失耦合构成，能够在多个时序尺度上学习判别性表征。随后，一个轻量的多尺度 Mamba 融合器通过基于 SSM 的聚合统一这些多尺度特征，从而实现精确的动作边界定位。MS-Temba 仅有 17M 参数，却在密集标注的 ADL 基准 TSU 与 Charades 上取得最优性能，并进一步泛化到长视频摘要任务，在 TVSum 与 SumMe 上刷新最优结果。
英文原文摘要 1297 字符
Temporal Action Detection (TAD) in untrimmed videos poses significant challenges, particularly for Activities of Daily Living (ADL) requiring models to (1) process long-duration videos, (2) capture temporal variations in actions, and (3) simultaneously detect dense overlapping actions. Existing CNN and Transformer-based approaches, struggle to jointly capture fine-grained detail and long-range structure at scale. State-space Model (SSM) based Mamba offers powerful long-range modeling, but naive application to TAD collapses fine-grained temporal structure and fails to account for the challenges inherent to TAD. To this end, we propose Multi-Scale Temporal Mamba (MS-Temba), which extends Mamba to TAD with newly introduced dilated SSMs. Each Temba block, comprising dilated SSMs coupled with our proposed additional losses, enables the learning of discriminative representations across temporal scales. A lightweight Multi-scale Mamba Fuser then unifies these multi-scale features via SSM-based aggregation, yielding precise action-boundary localization. With only 17M parameters, MS-Temba achieves state-of-the-art performance on densely labeled ADL benchmarks TSU & Charades, and further generalizes to long-form video summarization, setting new state-of-the-art results on TVSum & SumMe.