Finding Interesting Moments in Videos: Audio-visual and Unsupervised Video Highlight Detection

• Author / Creator

  Badamdorj, Taivanbat

• Video is the dominant medium through which we consume information. Due to the sheer volume of video content, it is of great interest in the research community to develop automated ways to make this content manageable. In video highlight detection, our goal is to automatically find interesting moments in videos. This makes it much easier to browse, search, categorize, and edit large amounts of video data. In this thesis, we develop two new state-of-the-art methods for video highlight detection. We first explore the use of audio for highlight detection. This is largely unexplored territory despite audio providing useful cues for finding interesting moments. Then we develop a new unsupervised method that alleviates the need for manual annotations that tell us exactly which parts of a video are interesting. Although our main focus is on video highlight detection, our models are general and we believe that they are applicable to other tasks such as action recognition, and video classification.

• Subjects / Keywords

  • highlight detection
  • audio-visual
  • unsupervised
  • machine learning
  • video highlight detection
  • video summarization

• Graduation date

  Spring 2022

• Type of Item

  Thesis

• Degree

  Master of Science

• DOI

  https://doi.org/10.7939/r3-1b4p-cr87

• License

  This thesis is made available by the University of Alberta Libraries with permission of the copyright owner solely for non-commercial purposes. This thesis, or any portion thereof, may not otherwise be copied or reproduced without the written consent of the copyright owner, except to the extent permitted by Canadian copyright law.