ChainDance:参数高效的可变规模群舞生成框架
Get the story
2026年10月9日,arXiv Computer Vision(一手)报道了ChainDance框架。该框架将群舞生成重构为Chain-of-Dancers序列分解,使单一模型无需重训即可适配可变人数并保留各舞者身份。其基于冻结的单人扩散模型骨干,新增RATE与GAME两个轻量模块,并在推理时采用免训练噪声优化保证空间协调。在AIOZ-GDance数据集上,ChainDance达到SOTA运动质量与群舞协调性,参数量减少3-4倍,训练时间缩短3-6倍。
Generated from reports · updated 3 hr ago
Timeline
Follow the coverage from different angles.
- arXiv · Computer VisionChainDance:突破群舞人数限制的参数高效群舞生成框架
ChainDance 将群舞生成重构为 Chain-of-Dancers 序列分解,使单一模型无需重训即可适配可变人数并保留各舞者身份。该框架基于冻结的单人扩散模型骨干,新增 RATE 与 GAME 两个轻量模块,并在推理时采用免训练噪声优化保证空间协调。在 AIOZ-GDance 上达到 SOTA 运动质量与群舞协调性,参数量减少 3-4 倍,训练时间缩短 3-6 倍。
Heat trend
Current heat 9·Comparable peak 10(Oct 9)·Comparable change over 24 hours –
The trend compares only the same participants observed continuously; its range may be smaller than the current heat count. Move or click on the chart to inspect hourly heat; use the left and right arrow keys to switch.