Annotating datasets for the computational analysis of film
David Bamman, Professor
Information, School of
Applications for Spring 2025 are closed for this project.
Methods in computer vision have reached a level of maturity that we can now develop computational instruments to measure a wide range of phenomena in film -- which actors are present in frame, the poses they have with respect to each other, the boundaries between shots -- which opens the door to a wide range of analytical questions. In this project, you will work on annotating phenomena in film clips in order to develop computational models to identify them. This can include the setting of scene (is it set in a kitchen, office, restaurant, outdoors?) or other aspects of film language (is this scene an establishing shot?). In your application, describe some aspect of film you would like to be able to measure, and why it would be interesting to do so.
Role: Tasks:
This project will involve reading research literature and watching short film clips to create annotated data for training and evaluation. Participation in biweekly group meetings to discuss progress and questions (lasting one hour) is required.
Qualifications: Background in film studies and/or a strong interest in film. No programming or computational experience is required.
Hours: 9-11 hrs
Off-Campus Research Site: Online
Engineering, Design & Technologies Arts & Humanities Digital Humanities and Data Science