A Survey on Vision-based Sign Language Recognition and Translation via Deep Learning

PhD Qualifying Examination


Title: "A Survey on Vision-based Sign Language Recognition and Translation via 
Deep Learning"

by

Mr. Zhe NIU


Abstract:

Sign language recognition and translation (SLR and SLT) aim to bridge the 
communication gap between the deaf and hearing people by transcribing or 
translating the sign video into text, which is a challenging task that involves 
the expertise in computer vision and neural language processing. Over past 
decades, hand-crafted features together with statistical sequence modeling 
methods have been widely used in SLR and SLT. With the rapid growth of the deep 
learning techniques, researchers have been switching from the legacy 
recognition and translation pipeline to neural network-based end-to-end 
systems, which have achieved superior performance to the legacy method. Despite 
this, current end-to-end SLR and SLT systems suffer from the generalizability 
issue and are not suitable for realistic scenarios. In this survey, we give a 
comprehensive introduction to the neural network-based SLR and SLT systems. 
Several spatial and sequential feature extraction network and sequence modeling 
techniques are introduced together with some recent related works. Potential 
research directions are pointed out in the end.


Date:			Thursday, 9 January 2020

Time:                  	3:00pm - 5:00pm

Venue:                  Room 3494
                         Lifts 25/26

Committee Members:	Dr. Brian Mak (Supervisor)
 			Prof. Dit-Yan Yeung (Chairperson)
 			Dr. Qifeng Chen
 			Dr. Yangqiu Song


**** ALL are Welcome ****