ICT-TV-Series Video Database

 

 

 

 

 

 

 

 

Summary

We construct two large scale video databases from two hit American shows, i.e., the Big Bang Theory (BBT), and Prison Break (PB). These two TV series are quite different in their filming styles: BBT is a sitcom about 20 minutes per episode with a main cast of 5 characters and mostly takes place indoors; on the other hand, PB has an average length of about 42 minutes per episode, where many shots are set outside, resulting in a large range of different illumination. To extract face tracks, several techniques, e.g., shot boundary detection, face detection, tracking, and facial landmark localization are involved. To guarantee the purity of databases, we also invite 5 fans of each TV series to annotate every extracted face track. More details, we deal with the whole first season of both TV series, i.e., 17 episodes for BBT, and 22 episodes for PB, and finally we collect 4,667 and 9,435 face tracks from BBT and PB, respectively.

 

Citation

This database is made available for research purposes within the scientific community. Any kind of commercial use of this database is forbidden. The following paper should be cited in any publication using this database:

 

[1] Yan Li, Ruiping Wang, Shiguang Shan, and Xilin Chen, “Hierarchical Hybrid Statistic based Video Binary Code and Its Application to Face Retrieval in TV-Series,” In 11th International Conference on Automatic Face and Gesture Recognition. IEEE, 2015.

​

Download

Please feel free to contact yan.li@vipl.ict.ac.cn or ruiping.wang@vipl.ict.ac.cn.