发表机构
School of Cyber Science and Engineering, Southeast University; School of Computer Science and Engineering, Southeast University; Purple Mountain Laboratories（东南大学网络空间安全学院; 东南大学计算机科学与工程学院; 紫金山实验室）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
GenPrior是首个利用预训练T2M模型生成先验的ZSAR框架，通过分散门控特征融合和生成原型细化，在NTU-60等数据集上实现零样本及广义零样本动作识别的SOTA性能。
AI 中文摘要
零样本骨骼-based动作识别（ZSAR）旨在通过对齐骨骼特征与文本语义来识别未见过的动作类别。然而，现有方法依赖于文本衍生的原型，这些原型固有地缺乏几何结构和物理约束，导致明显的语义-运动学差距。为弥合这一差距，我们提出GenPrior，这是首个利用预训练文本到动作（T2M）模型的生成先验用于ZSAR的框架。具体而言，我们引入分散门控特征融合，从生成的运动序列中提取运动学原型和类内分散，并采用学习到的门控网络将可靠的结构线索自适应注入文本嵌入，同时抑制合成伪影。此外，我们提出生成原型细化，利用这些生成增强的原型作为锚点挖掘高置信度的未见过样本，校准类原型以逼近真实分布，从而释放显著的性能提升。在NTU-60、NTU-120和PKU-MMD上的大量实验表明，GenPrior在零样本和广义零样本设置下均实现了最先进的性能。代码可在该https URL获取。
英文摘要
Zero-shot skeleton-based action recognition (ZSAR) aims to recognize unseen action categories by aligning skeleton features with textual semantics. However, existing methods rely on text-derived prototypes that inherently lack geometric structure and physical constraints, resulting in a pronounced \textit{semantic-kinematic gap}. To bridge this gap, we propose \textbf{GenPrior}, the first framework to exploit generative priors from pre-trained Text-to-Motion (T2M) models for ZSAR. Specifically, we introduce Dispersion-Gated Feature Fusion, which distills kinematic prototypes and intra-class dispersion from generative motion sequences and employs a learned gating network to adaptively inject reliable structural cues into textual embeddings while suppressing synthetic artifacts. Furthermore, we propose Generative Prototype Refinement, which leverages these generation-enhanced prototypes as anchors to mine high-confidence unseen samples, calibrating class prototypes toward the true distribution and thereby unleashing strong performance gains. Extensive experiments on NTU-60, NTU-120, and PKU-MMD demonstrate that GenPrior achieves state-of-the-art performance under both zero-shot and generalized zero-shot settings. Code is available at https://github.com/jidongkuang/GenPrior.
CommentsAccepted by ACMMM 2026
