Date of Award
Fall 11-13-2024
Degree Type
Dissertation
Degree Name
Doctor of Philosophy (PhD)
School
School of Computing
First Advisor
Daniela Stan Raicu, PhD
Second Advisor
Jacob Furst, PhD
Third Advisor
Thiru Ramaraj, PhD
Fourth Advisor
Yan Yan, PhD
Abstract
Performing video analysis for activity recognition presents challenges beyond classification, including obtaining class labels and performing temporal localization. One such challenge is precisely labeling a video with class labels having the exact start and end frames of an activity - a difficult task for a human to perform. Moreover, the task of annotating a video at any level of precision can quickly become tedious, impacting the attentiveness of the annotator and resulting in class label errors. Temporally localizing an activity within a video presents a second challenge. This dissertation investigates novel signal and image processing methods for motion features extracted in reference to the subject of the video. Our results show that shapes found in motion data (such as the speed of the subject) can provide an indication of where activities begin and end and have the potential to improve the class label refinement and temporal localization.
Recommended Citation
Piane, Jennifer, "Video label refinement and temporal localization using motion signal patterns" (2024). College of Computing and Digital Media Dissertations. 63.
https://via.library.depaul.edu/cdm_etd/63