Hand Gesture Recognition via Deep Learning

Citation

Ewe, Edmond Li Ren and Lee, Chin Poo and Kwek, Lee Chung (2022) Hand Gesture Recognition via Deep Learning. In: 2nd FET PG Engineering Colloquium Proceedings 2022, 1-15 December 2022, Multimedia University, Malaysia. (Unpublished)

[img] Text
07_Edmond Ewe_PPM colloquium Abstract_Second Half 2022.pdf
Restricted to Registered users only

Download (69kB)

Abstract

Background - Gesture recognition has been studied for a while within the fields of computer vision and pattern recognition. Gesture can be defined as a meaningful physical movement of the fingers, hands, arms, or other parts of the body with the purpose to convey information for the environment interaction. Vision-based hand gesture recognition is critical in its application however, there are challenges that will need to be overcome such as variations in the background, illuminations, hand orientation and size, and similarities among gestures. Traditional machine learning approaches has been widely used in visionbased hand gesture recognition for the past years but the complexity of its processing especially on the handcrafted feature extraction has been the major challenge. The effectiveness of the handcrafted feature extraction technique has not been proven across various datasets in comparison to deep learning techniques. Purpose – The purpose of the research is to investigate the limitations of the existing methods used in vision-based hand gesture recognition and propose deep learning approaches for vision-based hand gesture recognition. Design/methodology/approach – Publicly available dataset such as American Sign Language (ASL), ASL with digits and NUS dataset will be used in this study. The expected research outcomes are deep learning approaches with better performance in vision-based hand gesture recognition. Findings – Robustness of a feature extraction model is critical or key to having a good classification especially the ability to be used across different datasets. Deep learning model has proven its suitability in comparison to static inputed feature requirement. However, the usage of deep learning model in hand gesture recognition (HGR) has yet to reach its full potential, thus, various methodology for HGR can still be explored. Research limitations– The limitations in the field of deep learning is often gated by the dataset size and also the sufficiency of computation resources. Originality/value – The self developed hybrid model which utilizes CNN for feature extraction and machine learning method for classification.

Item Type: Conference or Workshop Item (Other)
Uncontrolled Keywords: Hand Gesture Recognition, Deep Learning, Hybrid Model
Subjects: Q Science > Q Science (General) > Q300-390 Cybernetics
Divisions: Faculty of Engineering and Technology (FET)
Depositing User: Ms Nurul Iqtiani Ahmad
Date Deposited: 15 Feb 2023 09:16
Last Modified: 15 Feb 2023 09:16
URII: http://shdl.mmu.edu.my/id/eprint/10728

Downloads

Downloads per month over past year

View ItemEdit (login required)