2026-09-18T22:58:37Zhttps://dspace.mit.edu/server/oai/requestoai:dspace.mit.edu:1721.1/1433472022-06-16T03:34:59Zcom_1721.1_7582com_1721.1_7581col_1721.1_131023
Berwick, Robert C.
Peng, Feifei
Pailet, Gregory
Massachusetts Institute of Technology. Department of Electrical Engineering and Computer Science
2022-06-15T13:14:13Z
2022-06-15T13:14:13Z
2022-02
2022-02-22T18:32:15.532Z
https://hdl.handle.net/1721.1/143347
In this thesis, we explore the task of generating highlight videos from sports games through the means of assessing the level of excitement of such videos to extract interesting moments from a game as well as utilize NLP techniques to generate captions for such videos. We create pipelines for the extraction of highlight clips using an audio heuristic for which we obtain transcriptions and, using a defined schema for exciting captions, fine-tune pre-trained transformer models to extract the best sentence from the video clip to use as a caption. Our results show improvements over baselines that solely use emotion-prediction categories of input sentences, suggesting our models are able to learn additional features to determine the excitement of captions.
M.Eng.
Massachusetts Institute of Technology
In Copyright - Educational Use Permitted
Copyright MIT
http://rightsstatements.org/page/InC-EDU/1.0/
Using Sports Videos to Showcase Exciting Content to Viewers
Thesis
application/pdf
Master
Master of Engineering in Electrical Engineering and Computer Science
Publication
unknown
unknown
<Publication xmlns="https://www.openaire.eu/cerif-profile/1.1/" id="6583e2e2-d88b-47f5-aa6d-c5c012d6ff79">
<Type xmlns="https://www.openaire.eu/cerif-profile/vocab/COAR_Publication_Types">http://purl.org/coar/resource_type/c_1843</Type>
<Title>Using Sports Videos to Showcase Exciting Content to Viewers</Title>
<PublishedIn>
<Publication>
</Publication>
</PublishedIn>
<PublicationDate>2022-02</PublicationDate>
<Authors>
<Author>
<DisplayName>Pailet, Gregory</DisplayName>
<Affiliation>
<OrgUnit>
</OrgUnit>
</Affiliation>
</Author>
</Authors>
<Editors>
</Editors>
<Publishers>
<Publisher>
<DisplayName>Massachusetts Institute of Technology</DisplayName>
<OrgUnit />
</Publisher>
</Publishers>
<License>http://rightsstatements.org/page/InC-EDU/1.0/</License>
<Abstract>In this thesis, we explore the task of generating highlight videos from sports games through the means of assessing the level of excitement of such videos to extract interesting moments from a game as well as utilize NLP techniques to generate captions for such videos. We create pipelines for the extraction of highlight clips using an audio heuristic for which we obtain transcriptions and, using a defined schema for exciting captions, fine-tune pre-trained transformer models to extract the best sentence from the video clip to use as a caption. Our results show improvements over baselines that solely use emotion-prediction categories of input sentences, suggesting our models are able to learn additional features to determine the excitement of captions.</Abstract>
<Access xmlns="http://purl.org/coar/access_right"
>
</Access>
</Publication>