Masip David, North Michael S, Todorov Alexander, Osherson Daniel N
Estudis d'Informatica Multimedia i Telecomunicacions, Universitat Oberta de Catalunya, Barcelona, Spain ; Computer Vision Center, Universitat Autonoma de Barcelona, Barcelona, Spain.
Department of Psychology, Columbia University, New York, New York, United States of America.
PLoS One. 2014 Feb 4;9(2):e87434. doi: 10.1371/journal.pone.0087434. eCollection 2014.
We introduce a computer vision problem from social cognition, namely, the automated detection of attitudes from a person's spontaneous facial expressions. To illustrate the challenges, we introduce two simple algorithms designed to predict observers' preferences between images (e.g., of celebrities) based on covert videos of the observers' faces. The two algorithms are almost as accurate as human judges performing the same task but nonetheless far from perfect. Our approach is to locate facial landmarks, then predict preference on the basis of their temporal dynamics. The database contains 768 videos involving four different kinds of preferences. We make it publically available.
我们引入一个来自社会认知的计算机视觉问题，即从人的自发面部表情中自动检测态度。为了说明其中的挑战，我们介绍两种简单算法，旨在根据观察者面部的隐蔽视频来预测观察者对图像（如名人图像）的偏好。这两种算法几乎与执行相同任务的人类评判员一样准确，但仍远非完美。我们的方法是定位面部特征点，然后根据其时间动态来预测偏好。该数据库包含768个涉及四种不同偏好的视频。我们将其公开。
