Calendar - University of Houston
Skip to main content

[Defense] Visual Summarization of Lecture Video Segments for Enhanced Navigation

Tuesday, September 15, 2020

2:30 pm - 4:00 pm

In Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy
Mohammad Rajiur Rahman
will defend his proposal
Visual Summarization of Lecture Video Segments for Enhanced Navigation


Abstract

Recorded lecture videos are gaining popularity every day. The traditional video format lacks features that enable quick access to specific items of interest within it. Multiple studies have shown strong demand amongst students for such features which ease their learning experience by introducing quick searchability, convenient review, and skimming through targeted portions of the video. My research is motivated to make lecture videos more accessible by adding enhanced navigation.

A real-world lecture video portal, ‘VideoPoints’, improves access to lecture videos through indexing, captioning, and search features. Indexing splits the video into segments each discussing a particular topic, thus, allowing learners to navigate through the subject matter easily. However, the segments are represented by the first frame which is generally not very descriptive of the material discussed. Instead, a visual summary frame containing important content from all the frames of the segment is expected to be more informative.

This thesis proposes a novel method to generate a visual summary of a lecture video segment by computing similarities between the images extracted from the segment and employing a graph-based algorithm to identify the subset of most representative images. The findings from this research are integrated into ‘VideoPoints’ portal. The result of an evaluation survey confirms that the methodology introduced in this work produces good quality visual summaries that are practically useful for lecture video navigation. I want to explore other methods to compute image similarities and improve the algorithm to generate better summaries. I will collect a larger body of ground truth with additional surveys to evaluate my methods.


 Tuesday, September 15, 2020
2:30PM - 4:00PM
Online via MS Teams (click link)

Dr. Jaspal Subhlok, dissertation advisor

Faculty, students and the general public are invited.

Location
Online via MS Teams
Contact
csevents@uh.edu