Repository of televised news, including (for many) captions and rough statistics for content
Dataset of 8 million annotated YouTube videos, including a variety of audio and visual features.
Dataset of over 13,000 images of faces (labeled with names) taken from the internet, including over 1,600 people with multiple pictures
Datasets for training affect recognition and for perception studies
Video data from head-mounted camera (first person or egocentric perspective) during activities at a theme park