Search engine for discovering works of Art, research articles, and books related to Art and Culture
Javascript must be enabled to continue!
Cross-Age Face Verification Using Generative Adversarial Networks (GAN) with Landmark Feature
View through CrossRef
Cross-age face verification is a complex problem in biometric recognition in terms of aging, a naturally changing face structure, and face landmark configuration changes over time. In this paper, a new cross-age face verification method is proposed with a Generative Adversarial Network (GAN) and a mix of landmark-based features. Realistic aging of a face with identity-specific landmarks, such as eyes, nose, and mouth, is generated for effective face recognition in a range of age groups. Performance testing with an in-house collected face dataset of 200 face images exhibited effectiveness in changing face configuration and face shape transformations, such as a fuller face thinning and thin face becoming fuller. Comparison with direct face verification showed increased values of similarity, such as 32.57% to 63.80%, reduced values of feature distance, such as 0.6743 to 0.3620, and improvement in accuracy for the ArcFace, VGG-Face, and Facenet architectures. ArcFace exhibited an improvement in accuracy with an increase in value from 82.64% to 86.02%, VGG-Face with an improvement in value from 76.23% to 80.57%, and Facenet with an improvement in value from 67.54% to 74.48%. These observations validate the effectiveness of the proposed method in overcoming age-related complications and improving cross-age face verification performance. In future work, we plan to investigate a larger dataset and model refinement to realize performance improvement and real-life biometric suitability.
ASCEE Publications
Title: Cross-Age Face Verification Using Generative Adversarial Networks (GAN) with Landmark Feature
Description:
Cross-age face verification is a complex problem in biometric recognition in terms of aging, a naturally changing face structure, and face landmark configuration changes over time.
In this paper, a new cross-age face verification method is proposed with a Generative Adversarial Network (GAN) and a mix of landmark-based features.
Realistic aging of a face with identity-specific landmarks, such as eyes, nose, and mouth, is generated for effective face recognition in a range of age groups.
Performance testing with an in-house collected face dataset of 200 face images exhibited effectiveness in changing face configuration and face shape transformations, such as a fuller face thinning and thin face becoming fuller.
Comparison with direct face verification showed increased values of similarity, such as 32.
57% to 63.
80%, reduced values of feature distance, such as 0.
6743 to 0.
3620, and improvement in accuracy for the ArcFace, VGG-Face, and Facenet architectures.
ArcFace exhibited an improvement in accuracy with an increase in value from 82.
64% to 86.
02%, VGG-Face with an improvement in value from 76.
23% to 80.
57%, and Facenet with an improvement in value from 67.
54% to 74.
48%.
These observations validate the effectiveness of the proposed method in overcoming age-related complications and improving cross-age face verification performance.
In future work, we plan to investigate a larger dataset and model refinement to realize performance improvement and real-life biometric suitability.
Related Results
Highmobility AlGaN/GaN high electronic mobility transistors on GaN homo-substrates
Highmobility AlGaN/GaN high electronic mobility transistors on GaN homo-substrates
Gallium nitride (GaN) has great potential applications in high-power and high-frequency electrical devices due to its superior physical properties.High dislocation density of GaN g...
Studies on the Influences of i-GaN, n-GaN, p-GaN and InGaN Cap Layers in AlGaN/GaN High-Electron-Mobility Transistors
Studies on the Influences of i-GaN, n-GaN, p-GaN and InGaN Cap Layers in AlGaN/GaN High-Electron-Mobility Transistors
Systematic studies were performed on the influence of different cap layers of i-GaN, n-GaN, p-GaN and InGaN on AlGaN/GaN high-electron-mobility transistors (HEMTs) grown on sapphi...
GGADN: Guided Generative Adversarial Dehazing Network
GGADN: Guided Generative Adversarial Dehazing Network
Abstract
Image dehazing has always been a challenging topic in image processing. The development of deep learning methods, especially the Generative Adversarial Networks(GA...
MSG-Point-GAN: Multi-Scale Gradient Point GAN for Point Cloud Generation
MSG-Point-GAN: Multi-Scale Gradient Point GAN for Point Cloud Generation
The generative adversarial network (GAN) has recently emerged as a promising generative model. Its application in the image field has been extensive, but there has been little rese...
Single image super-resolution using capsule generative adversarial network
Single image super-resolution using capsule generative adversarial network
The current research aims to investigate and propose a Generative Adversarial Network (GAN) architecture [53] using capsule network architecture [76] in the discriminator module of...
Research on Style Migration Techniques Based on Generative Adversarial Networks in Chinese Painting Creation
Research on Style Migration Techniques Based on Generative Adversarial Networks in Chinese Painting Creation
Abstract
The continuous progress and development of science and technology have brought rich and diverse artistic experiences to the current society. The image style...
Shenzi 16-Inch Oil Export SCR CVA Verification
Shenzi 16-Inch Oil Export SCR CVA Verification
Abstract
In 2006 Enterprise developed a 16-inch oil export system from Shenzi field located in Green Canyon Block 653 in the Gulf of Mexico, approximately 120 nau...
Enhancing Adversarial Robustness through Stable Adversarial Training
Enhancing Adversarial Robustness through Stable Adversarial Training
Deep neural network models are vulnerable to attacks from adversarial methods, such as gradient attacks. Evening small perturbations can cause significant differences in their pred...