Spatio-Temporal Deep Learning for Robust and Efficient Face Presentation Attack Detection
Dates: 2024
Principal Investigator: Dr. Shujaat Khan
Description: The proposed research aims to contribute to
Saudi Vision 2030 by pioneering an innovative approach to face anti-spoofing
using deep learning techniques. Over 10 months, the project will focus on
leveraging temporal depth information in video sequences to enhance model
accuracy, a facet previously underexplored in anti-spoofing. Addressing the
evolving threat landscape in the digital age, the research aligns with the
Vision's goal of technological advancement and a secure digital environment. By
embracing temporal cues overlooked in traditional methods, the proposed model
aims to discern genuine users from spoofing attempts with unprecedented
accuracy, as evidenced by promising preliminary results. This research directly
addresses the critical need for robust identity verification systems in an
increasingly digitized world, offering a defense against spoofing attacks and
instilling confidence in online interactions.