arXiv:2604.18274cs.CV2026-04被引 1

用类液态神经机制实现高效动作检测,参数少一半还能提速。

LiquidTAD: Efficient Temporal Action Detection via Parallel Liquid-Inspired Temporal Relaxation

论文配图:LiquidTAD: Efficient Temporal Action Detection via Parallel Liquid-Inspired Temporal Relaxation
图 1 · 摘自论文原文
  • 用并行化非递归公式替代传统时序求解,支持硬件通用部署。
  • 在THUMOS-14上达69.46% mAP,参数仅1082万,减少60%以上。
  • 适合资源受限场景,如移动端或边缘设备部署动作检测。

时序动作检测(TAD)需要在长视频中精确定位动作边界。现有高性能方法虽准确度高,但参数量大、计算开销高,且依赖专用算子,难以跨平台部署。本文提出LiquidTAD,将类液态神经动力学的指数衰减先验提炼为并行时序算子,而非复现完整液态神经网络(LNN)动态。通过引入并行类液态松弛机制,完全避免序列微分方程求解,采用标准神经运算实现全向量化非递归结构,实现与时间长度线性相关的复杂度,支持硬件无关部署。同时,层级衰减率共享策略在多尺度特征金字塔间自适应调整松弛先验,稳定优化并隐式补偿深层特征的时间压缩。在THUMOS-14和ActivityNet-1.3上的实验表明,LiquidTAD在保持强准确率的同时显著降低模型规模。具体地,在THUMOS-14上达到69.46%平均mAP,仅需10.82M参数和27.17G FLOPs,相比ActionFormer参数量减少超60%。

原文摘要 · Abstract (English)

Temporal Action Detection (TAD) requires precise localization of action boundaries within long, untrimmed video sequences. While current high-performing methods achieve strong accuracy, they are often characterized by excessive parameter counts, substantial computational overhead, and a reliance on specialized operators that hinder deployment across diverse hardware platforms. This paper presents LiquidTAD, a framework that distills the exponential relaxation prior of liquid neural dynamics into a parallel temporal operator, rather than reproducing full Liquid Neural Network (LNN) dynamics. By introducing a Parallel Liquid-inspired Relaxation mechanism, sequential ODE solving is avoided through a fully vectorized, non-recursive formulation built entirely upon standard neural operations, enabling hardware-agnostic deployment with linear complexity with respect to the temporal length. A complementary Hierarchical Decay-Rate Sharing Strategy further adapts this relaxation prior across feature pyramid levels, stabilizing optimization and implicitly compensating for temporal compression in deeper layers. Experimental evaluations on THUMOS-14 and ActivityNet-1.3 demonstrate that LiquidTAD achieves accuracy competitive with strong baselines while substantially lowering the model footprint. Specifically, on THUMOS-14, LiquidTAD achieves 69.46\% average mAP with only 10.82M parameters and 27.17G FLOPs, reducing the parameter count by over 60\% compared with ActionFormer.

动作检测轻量化时序建模

Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品;中文卡片由大模型生成,请以原文为准。