Deep Learning Based Face Recognition with Sparse Representation Classification
- Publisher:
- SPRINGER INTERNATIONAL PUBLISHING AG
- Publication Type:
- Conference Proceeding
- Citation:
- Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 2017, 10636 LNCS, pp. 665-674
- Issue Date:
- 2017-01-01
Open Access
Copyright Clearance Process
- Recently Added
- In Progress
- Open Access
This item is open access.
Feature extraction is an essential step in solving real-world pattern recognition and classification problems. The accuracy of face recognition highly depends on the extracted features to represent a face. The traditional algorithms uses geometric techniques, comprising feature values including distance and angle between geometric points (eyes corners, mouth extremities, and nostrils). These features are sensitive to the elements such as illumination, variation of poses, various expressions, to mention a few. Recently, deep learning techniques have been very effective for feature extraction, and deep features have considerable tolerance for various conditions and unconstrained environment. This paper proposes a two layer deep convolutional neural network (CNN) for face feature extraction and applied sparse representation for face identification. The sparsity and selectivity of deep features can strengthen sparseness for the solution of sparse representation, which generally improves the recognition rate. The proposed method outperforms other feature extraction and classification methods in terms of recognition accuracy.
Please use this identifier to cite or link to this item: