Low complexity hybrid holistic – landmark based approach for face recognition

Modern Face Recognition (FR) systems have to cope with high intra-class variability. Face is a 3D object with wide variability in appearances from each pose. This, along with other causes like luminance variation, becomes one of the main reasons for the high variability. While incremental learning i...

Full description

Saved in:
Bibliographic Details
Published in:Multimedia tools and applications 2021-08, Vol.80 (20), p.30199-30212
Main Authors: Asad, Muhammad, Hussain, Ayyaz, Mir, Usama
Format: Article
Language:English
Subjects:
Classification
Complexity
Computer Communication Networks
Computer Science
Computing costs
Data Structures and Information Theory
Deep learning
Distance learning
Face recognition
Facial recognition technology
Multimedia
Multimedia Information Systems
Special Purpose and Application-Based Systems
Variability
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Modern Face Recognition (FR) systems have to cope with high intra-class variability. Face is a 3D object with wide variability in appearances from each pose. This, along with other causes like luminance variation, becomes one of the main reasons for the high variability. While incremental learning is an unsuited approach for most scenarios, techniques proposed to deal with this issue are often of enormous complexity during training phase and require computational power during operation that is scarcely credible for compact devices. This paper proposes a novel technique where pose specific classification system has been used to attain better classification with low computational cost. State of the art performance measures have been used to assess the validity. Results demonstrate the effectiveness of our pose specific classification system.
ISSN:1380-7501
1573-7721
DOI:10.1007/s11042-020-08872-z