More about HKUST
Towards Temporally Consistent Video Editing
The Hong Kong University of Science and Technology
Department of Computer Science and Engineering
PhD Thesis Defence
Title: "Towards Temporally Consistent Video Editing"
By
Mr. Hao OUYANG
Abstract:
The ubiquity of video content in our daily lives has catalyzed a corresponding
surge in demand for advanced video editing techniques. The main goal of video
editing is the transformation of visual data into outputs that are not only
visually pleasing but also temporally consistent. Such algorithms hold
significant potential for a myriad of practical applications, enabling users to
seamlessly remove unwanted objects, rectify facial anomalies, and generate
aesthetically pleasing outcomes. Despite considerable advancements in video
editing methodologies, persistent issues remain, notably the emergence of
abrupt color artifacts and severe flickering, which detract from the realism of
edited videos.
This thesis is committed to enhancing video editing performance with a special
focus on temporal consistency. Our approach is tripartite: Firstly, we employ
deep internal learning to tap into the intrinsic structures of video data,
aiming to achieve improved consistency. Secondly, we integrate 3D information
to further stabilize temporal elements. Lastly, we refine video representation
through a content deformation field, which adapts existing image algorithms for
application to video sequences. These diverse strategies have been tested
across a variety of editing tasks and have demonstrated commendable efficacy in
improving the fidelity and coherence of video editing processes.
Date: Monday, 6 May 2024
Time: 4:00pm - 6:00pm
Venue: Room 5506
Lifts 25/26
Chairman: Prof. Yi-Kuen LEE (MAE)
Committee Members: Prof. Qifeng CHEN (Supervisor)
Prof. Chi-Keung TANG
Prof. Pedro SANDER
Prof. Ling SHI
Prof. Rynson LAW (CityU)