Face Video DataBase

版权声明:本文为博主原创文章,未经博主允许不得转载。 https://blog.csdn.net/LEE18254290736/article/details/88972482

官方地址

https://vdb.kyb.tuebingen.mpg.de/login.php

介绍:

Overview

This database contains videos of facial action units which were recorded starting in the autumn of 2003 at the MPI for Biological Cybernetics in the Face and Object Recognition Group, department Prof. Bülthoff, using the Videolab facilities created by Mario Kleiner and Christian Wallraven.

Technology

The Videolab technology allows us to record facial movements from six different viewpoints at the same time while maintaining a very high synchronization between the different cameras. Please refer to the Technical Report 123 by M. Kleiner et al. for a detailed description of the recording technology or the Videolab description for an overview.

Recording Setup

Here is a brief description of the camera arrangment used for the recordings:

The video cameras were arranged in a semi-circle around the subject at a distance of roughly 1.3m, as shown in Fig. 1 Each camera was centered on the subject and leveled. The cameras recorded 25 frames/sec in full PAL video resolution (786*576, non-interlaced). In order to facilitate the recovery of rigid head motion, the subject wore a headplate with 6 green foam markers attached to it.

Camera Layout

Fig. 1 - Top view of camera layout used for recordings

Each recording contains one isolated action unit, repeated three times, with a (close to) neutral expression inbetween. For each action unit, there are six video files (one for each camera). Each video file has identical length and starts at exactly the same time. The videos were converted from raw single chip CCD data to RGB using a Bayer filter, then encoded as MPEG1 using mpeg2enc.

Publications

下载地址:

https://fgk.pw/i/fmT6jGe1155

猜你喜欢

转载自blog.csdn.net/LEE18254290736/article/details/88972482