FEAFA
收藏资源简介:
FEAFA数据集是由中国科学院工程科学学院创建,包含122名参与者在真实世界条件下的123个视频,总计99,356帧。该数据集通过自开发的Expression Quantification Tool进行详细标注,涵盖9个对称FACS动作单元、10个非对称FACS动作单元、2个对称FACS动作描述符和2个非对称FACS动作描述符,每个动作单元或描述符的强度以0到1的浮点数表示。FEAFA数据集主要用于面部表情分析和3D面部动画,旨在通过精确的动作单元值回归,推动面部动画技术的发展。
The FEAFA dataset was created by the School of Engineering Science, Chinese Academy of Sciences. It contains 123 videos from 122 participants captured under real-world conditions, with a total of 99,356 frames. This dataset was thoroughly annotated using its self-developed Expression Quantification Tool, covering 9 symmetric FACS action units, 10 asymmetric FACS action units, 2 symmetric FACS action descriptors, and 2 asymmetric FACS action descriptors. The intensity of each action unit or descriptor is represented as a floating-point number ranging from 0 to 1. The FEAFA dataset is primarily used for facial expression analysis and 3D facial animation, aiming to advance the development of facial animation technologies through precise regression of action unit values.

- 1FEAFA: A Well-Annotated Dataset for Facial Expression Analysis and 3D Facial Animation中国科学院工程科学学院 · 2019年



