Novel Deep Learning-Based Facial Forgery Detection for Effective Biometric Recognition DOI Creative Commons
Han‐Soo Kim

Applied Sciences, Год журнала: 2025, Номер 15(7), С. 3613 - 3613

Опубликована: Март 26, 2025

Advancements in science, technology, and computer engineering have significantly influenced biometric identification systems, particularly facial recognition. However, these systems are increasingly vulnerable to sophisticated forgery techniques. This study presents a novel deep learning framework optimized for texture analysis detect forgeries effectively. The proposed method leverages high-frequency features, such as roughness, color variation, randomness, which more challenging replicate than specific features. network employs shallow architecture with wide feature maps enhance efficiency precision. Furthermore, binary classification approach combined supervised contrastive addresses data imbalance strengthens generalization capabilities. Experimental results, conducted on three benchmark datasets (CASIA-FASD, CelebA-Spoof, NIA-ILD), demonstrate the model’s robustness, achieving an Average Classification Error Rate (ACER) of approximately 0.06, outperforming existing methods. ensures practical applicability real-time providing reliable efficient solution detection.

Язык: Английский

Thin Film Thickness Analysis Using a Deep Learning Algorithm with a Consideration of Reflectance Fluctuation DOI Creative Commons
Joonyoung Lee, Jonghan Jin

International Journal of Precision Engineering and Manufacturing-Smart Technology, Год журнала: 2025, Номер 3(1), С. 31 - 38

Опубликована: Янв. 1, 2025

A deep learning algorithm for thin film thickness analysis based on spectral reflectometry, using a dataset that reflects experimental conditions, has been proposed and implemented. This study extends our previous research, in which we designed an artificial neural network (ANN) theoretical reflectance spectrum datasets quantitatively evaluated it according to the international standard traceability system. The evaluation results indicated one of major sources uncertainty was offset between outputs ANN certified values reference materials (CRMs). In this study, focused how much factor related is affected by conditions instead datasets. By applying fluctuations obtained from experiments spectrum, created train under same as studies comparison. As result, improved about 30%. demonstrates importance having accurately reflect real-world training algorithms.

Язык: Английский

Процитировано

0
Novel Deep Learning-Based Facial Forgery Detection for Effective Biometric Recognition DOI Creative Commons
Han‐Soo Kim

Applied Sciences, Год журнала: 2025, Номер 15(7), С. 3613 - 3613

Опубликована: Март 26, 2025

Advancements in science, technology, and computer engineering have significantly influenced biometric identification systems, particularly facial recognition. However, these systems are increasingly vulnerable to sophisticated forgery techniques. This study presents a novel deep learning framework optimized for texture analysis detect forgeries effectively. The proposed method leverages high-frequency features, such as roughness, color variation, randomness, which more challenging replicate than specific features. network employs shallow architecture with wide feature maps enhance efficiency precision. Furthermore, binary classification approach combined supervised contrastive addresses data imbalance strengthens generalization capabilities. Experimental results, conducted on three benchmark datasets (CASIA-FASD, CelebA-Spoof, NIA-ILD), demonstrate the model’s robustness, achieving an Average Classification Error Rate (ACER) of approximately 0.06, outperforming existing methods. ensures practical applicability real-time providing reliable efficient solution detection.

Язык: Английский

Процитировано

0