Skip to main content



Master Thesis Defense: Yiran Shen

August 20, 2019

Speaker: Yiran Shen

Supervisors: Drs. S. Mokhov, J. Paquet

Examining Committee:
Drs. N. Shiri, J. Yang, A. Hanna (Chair)

Title: Toward a Flexible Facial Analysis Framework in OpenISS for Visual Effects

Date:Tuesday, August 20, 2019

Time: 13:00

Place: EV 3.309


Facial analysis, including tasks such as face detection, facial landmark detection, and facial expression recognition, is a significant research domain in computer vision for visual effects. It can be used in various domains such as facial feature mapping for movie animation, biometrics/face recognition for security systems, and driver fatigue monitoring for transportation safety assistance. Most applications involve basic face and landmark detection as preliminary analysis approaches before proceeding into further specialized processing applications. As technology develops, there are plenty of implementations and resources for each task available for researchers, but the key missing properties among them all are flexibility and usability. The integration of functionality components involves complex configurations for each connection joint which is typically problematic with poor reusability and adjustability. The lack of support for integrating different functionality components greatly impact the research effort and cost for individual researchers, which also leads us to the idea of providing a framework solution that can help regarding the issue once and for all. To address this problem, we propose a user-friendly and highly expandable facial analysis framework solution. It contains a core that supports fundamental services for the framework, and a facial analysis module composed of implementations for facial analysis tasks. We evaluate our framework solution and achieve our goals of instantiating the facial analysis specialized framework, which essentially perform tasks in face detection, facial landmark detection, and facial expression recognition. This framework solution as a whole, solves the industry problem of lacking an execution platform for integrated facial analysis implementations and fills the gap in visual effects industry.

Back to top Back to top

© Concordia University