论文解读 · 作者已确认

InterSyn: Interleaved Learning for Dynamic Motion Synthesis in the Wild

作者: Yiyi Ma, , Xiu Li, Chi Zhang, Xuelong Li

IEEE/CVF International Conference on Computer Vision (ICCV 2025) · 2025 · 页 12832-12841

3D human motionmulti-person interactioninterleaved learningmotion synthesiscoordination refinement

论文信息

作者
Yiyi Ma, Yuanzhi Liang, Xiu Li, Chi Zhang, and Xuelong Li
推荐论文引用
Yiyi Ma, Yuanzhi Liang, Xiu Li, Chi Zhang, and Xuelong Li. “InterSyn: Interleaved Learning for Dynamic Motion Synthesis in the Wild.” 2025 IEEE/CVF International Conference on Computer Vision (ICCV) (2025), 12832-12841. https://doi.org/10.1109/ICCV51701.2025.01192.

版本日期

arXiv 首次提交
2025-08-14
arXiv 最近修订
2025-08-14
来源核验日期
2026-07-31
核验状态
作者已于 2026-07-31 核验

论文概要

InterSyn 不把单人动作与多人交互视为彼此独立的任务:INS 从第一人称交互视角联合合成两类行为,REC 再细化角色之间的相对协调与同步。

研究问题

若将 solo motion 和 multi-person interaction 分开建模,模型难以学习个体行为与相互协调在真实动态交互中如何共同形成。

论文贡献

  • 对整合后的单人和多人动作执行 interleaved learning。
  • 提出 Interleaved Interaction Synthesis(INS),从第一人称视角统一建模 solo 与 interactive behavior。
  • 提出 Relative Coordination Refinement(REC),细化角色间的 mutual dynamics 与同步。

证据与评测范围

论文报告相比近期方法具有更高的 text-to-motion alignment 和更好的 diversity;该结论限定于已评测动作合成设置,准确数据集、baseline 和数值应引用 ICCV 论文。

适用范围与局限

已展示范围是利用整合后的单人和多人数据进行动作合成。摘要并未证明方法可泛化到任意人数、环境或评测分布外的交互语义。

Related work 定位

InterSyn 是强调 integrated learning 与 relative coordination 的多人交互动作方法,不应误写成“联合学习人物动作和动态环境”;这里的 dynamic 指单人及交互角色的运动动态。

Related Work 表述

Ma 等提出 InterSyn,通过 interaction-synthesis 模块交错学习单人和多人动作,并细化相对协调,以改善同步且与文本对齐的交互动作生成。

这是一段用于说明论文定位的简洁中性表述。

论文官方英文摘要

We present Interleaved Learning for Motion Synthesis (InterSyn), a novel framework that targets the generation of realistic interaction motions by learning from integrated motions that consider both solo and multi-person dynamics. Unlike previous methods that treat these components separately, InterSyn employs an interleaved learning strategy to capture the natural, dynamic interactions and nuanced coordination inherent in real-world scenarios. Our framework comprises two key modules: the Interleaved Interaction Synthesis (INS) module, which jointly models solo and interactive behaviors in a unified paradigm from a first-person perspective to support multiple character interactions, and the Relative Coordination Refinement (REC) module, which refines mutual dynamics and ensures synchronized motions among characters. Experimental results show that the motion sequences generated by InterSyn exhibit higher text-to-motion alignment and improved diversity compared with recent methods, setting a new benchmark for robust and natural motion synthesis. Additionally, our code will be open-sourced in the future to promote further research and development in this area.

摘要仅用于学术识别,版权仍归论文作者或出版方所有,不属于本页 CC BY 许可范围。

依据与出处

核对内容论文中的位置
问题陈述Abstract
方法与贡献Abstract; INS and REC method sections
评测结论Abstract; text-to-motion alignment and diversity experiments

主要核验来源: IEEE version of record (ICCV 2025 version of record; CVF open-access copy and arXiv:2508.10297v1 cross-checked).

如何引用

科研结论应引用论文本身;只有在复用本站原创解读时才引用本页。

Yiyi Ma, Yuanzhi Liang, Xiu Li, Chi Zhang, and Xuelong Li. “InterSyn: Interleaved Learning for Dynamic Motion Synthesis in the Wild.” 2025 IEEE/CVF International Conference on Computer Vision (ICCV) (2025), 12832-12841. https://doi.org/10.1109/ICCV51701.2025.01192.

复用许可

本页原创解读采用 CC BY 4.0:复用时须署名并链接本页。论文标题、摘要、图表和书目信息不在此许可范围内,仍保留原有权利。 Creative Commons Attribution 4.0 International.

一手来源与资源