Scholarlink

<p>We address the challenging problem of fine-grained text-driven human motion generation. Existing works generate imprecise motions that fail to accurately capture relationships specified in text due to: (1) lack of effective text parsing for detailed semantic cues regarding body parts, (2) not fully modeling linguistic structures between words to comprehend text comprehensively. To tackle these limitations, we propose a novel fine-grained framework Fg-T2M++ that consists of: (1) an <i>LLMs semantic parsing module</i> to extract body part descriptions and semantics from text, (2) a <i>hyperbolic text representation module</i> to encode relational information between text units by embedding the syntactic dependency graph into hyperbolic space, and (3) a <i>multi-modal fusion module</i> to hierarchically fuse text and motion features. Extensive experiments on HumanML3D and KIT-ML datasets demonstrate that Fg-T2M++ outperforms SOTA methods, validating its ability to accurately generate motions adhering to comprehensive text semantics.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Fg-T2M++: LLMs-Augmented Fine-Grained Text Driven Human Motion Generation

  • 期刊论文

  • 发表日期：

    2025年2月27日

查看全文
  • Yin Wang,
  • Mu Li,
  • Jiapeng Liu,
  • Zhiying Leng,
  • Frederick W. B. Li,
  • Ziyao Zhang,
  • Xiaohui Liang

摘要

We address the challenging problem of fine-grained text-driven human motion generation. Existing works generate imprecise motions that fail to accurately capture relationships specified in text due to: (1) lack of effective text parsing for detailed semantic cues regarding body parts, (2) not fully modeling linguistic structures between words to comprehend text comprehensively. To tackle these limitations, we propose a novel fine-grained framework Fg-T2M++ that consists of: (1) an LLMs semantic parsing module to extract body part descriptions and semantics from text, (2) a hyperbolic text representation module to encode relational information between text units by embedding the syntactic dependency graph into hyperbolic space, and (3) a multi-modal fusion module to hierarchically fuse text and motion features. Extensive experiments on HumanML3D and KIT-ML datasets demonstrate that Fg-T2M++ outperforms SOTA methods, validating its ability to accurately generate motions adhering to comprehensive text semantics.