MoDi:基于多种数据的无条件运动合成
MoDi: Unconditional Motion Synthesis from Diverse Data
摘要:
神经网络的出现彻底改变了运动合成领域。然而,学习从给定的分布无条件地合成运动仍然具有挑战性,特别是当运动是高度多样化的。在这项工作中,我们提出了MoDi -一种在无监督环境中从极其多样化,非结构化和无标签数据集训练的生成模型。在推理过程中,MoDi可以合成高质量,多样化的运动。尽管数据集中缺乏任何结构,但我们的模型产生了一个行为良好且高度结构化的潜在空间,该空间可以在语义上聚类,构成了一个强运动先验,有利于各种应用,包括语义编辑和人群模拟。此外,我们提出了一个编码器,将真实的运动转化为MoDi的自然运动流形,解决了各种不适定的挑战,如完成前缀和空间编辑。我们的定性和定量实验取得了最先进的结果,优于最近的SOTA技术。代码和训练模型可在https://sigal-raab.github.io/MoDi上获得。
英文摘要:
The emergence of neural networks has revolutionized the field of motion synthesis. Yet, learning to unconditionally synthesize motions from a given distribution remains challenging, especially when the motions are highly diverse. In this work, we present MoDi -- a generative model trained in an unsupervised setting from an extremely diverse, unstructured and unlabeled dataset. During inference, MoDi can synthesize high-quality, diverse motions. Despite the lack of any structure in the dataset, our model yields a well-behaved and highly structured latent space, which can be semantically clustered, constituting a strong motion prior that facilitates various applications including semantic editing and crowd simulation. In addition, we present an encoder that inverts real motions into MoDi's natural motion manifold, issuing solutions to various ill-posed challenges such as completion from prefix and spatial editing. Our qualitative and quantitative experiments achieve state-of-the-art results that outperform recent SOTA techniques. Code and trained models are available at https://sigal-raab.github.io/MoDi.
机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。