How to browse through my large video data

Due to the rise of lifelog cameras, we have personal video data that is too large to be watched. Video indexing has the potential to provide meta-information for faster video search. This work aims to support lifelog video indexing through automated face priority rating. In a user study, we identified parameters that allow for rating the importance of persons in a video. We implemented these findings to automatically predict the person's importance in video. We show that our algorithm predicts similar person priority ratings like the participants had given. Hence, we contribute to video-based lifelogging through indicating, implementing, and testing face indexing rules that predict how important a person in a video is perceived. Our findings can help to build video players that support users navigating through their large video data and reviewing sequences that recall important moments of life.