X2C
收藏资源简介:
X2C数据集是专门为真实人形机器人面部表情模仿而设计的一个高质量、高多样性、大规模的数据集。它包含10万个(图像、控制值)对,每张图像描绘了一个具有多样化微表情的人形机器人,并使用30个控制值对其进行注释,以表示真实表情配置。数据集的创建过程涉及从人形机器人执行面部表情动画的视频中提取图像,并由志愿者进行手动编辑和注释。X2C旨在解决现有数据集规模小、多样性不足、标注维度低的问题,为研究人形机器人面部表情模仿提供了新的资源。
The X2C dataset is a high-quality, large-scale and highly diverse dataset specifically developed for facial expression imitation of real humanoid robots. It consists of 100,000 (image, control value) pairs. Each image captures a humanoid robot displaying diverse micro-expressions, and is annotated with 30 control values to characterize the real facial expression configuration. Dataset construction involves extracting frames from videos of humanoid robots executing facial expression animations, followed by manual editing and annotation conducted by volunteers. X2C aims to address the shortcomings of existing datasets, such as small scale, insufficient diversity and low annotation dimensionality, providing a novel resource for research on humanoid robot facial expression imitation.
- 1X2C: A Dataset Featuring Nuanced Facial Expressions for Realistic Humanoid Imitation麦考瑞大学, 中山大学 · 2025年



