“Grounding Natural Language Phrases in Images and Video”
Monday, May 14, 2018
Seminar 4:00-5:00pm; Networking reception, 5:00-5:30pm
Hariri Institute for Computing, Room 180
Boston, MA 02215
Grounding Natural Language Phrases in Images and Video
Bryan Plummer, Postdoctoral Associate in Computer Science
Abstract: Grounding language in images has shown it help improve performance on many image-language tasks. To address this task, this talk will introduce an approach which learns a set of models, each of which capture a different concept which is useful in the task. These concepts can be predefined, such as attributes gleamed from the adjectives, as well as those which are automatically learned in a single-end-to-end neural network. He will also briefly address the more challenging detection style task, where the goal is to localize a phrase and determine if it is associated with an image.
Bio: Bryan Plummer completed his PhD at the University of Illinois at Urbana-Champaign advised by Svetlana Lazebnik. His thesis research has focused on the joint learning of images and text, with a general interest in visual recognition, representation learning, and computer vision for fashion.