An effective facial spoofing detection approach based on weighted deep ensemble learning

被引:0
作者
My Abdelouahed Sabri
Assia Ennouni
Abdellah Aarab
机构
[1] University Sidi Mohamed Ben Abdallah,Computer Science Department, Faculty of Sciences
[2] University Sidi Mohamed Ben Abdallah,Physics Department, Faculty of Sciences
来源
Signal, Image and Video Processing | 2024年 / 18卷
关键词
Spoofing detection; Deep learning; Ensemble learning; Face recognition; ROSE-Youtu dataset;
D O I
暂无
中图分类号
学科分类号
摘要
Deep learning has seen successful implementation in various domains, such as natural language processing, image classification, and object detection in recent times. In the field of biometrics, deep learning has also been used to develop effective anti-spoofing systems. Facial spoofing, the act of presenting fake facial information to deceive a biometric system, poses a significant threat to the security of face recognition systems. To address this challenge, we propose, in this paper, an effective and robust facial spoofing detection approach based on weighted deep ensemble learning. Our method combines the strengths of two powerful deep learning architectures, DenseNet201 and MiniVGG. The choice of these two architectures is based on a comparative study between DenseNet201, DenseNet169, VGG16, MiniVGG, and ResNet50, where DenseNet201 and MiniVGG obtained the best recall and precision scores, respectively. Our proposed weighted voting ensemble leverages each architecture-specific capabilities to make the final prediction. We assign weights to each classification model based on its performance, which are determined by a mathematical formulation considering the trade-off between recall and precision. To validate the effectiveness of our proposed approach, we evaluate it on the challenging ROSE-Youtu face liveness detection dataset. Our experimental results demonstrate that our proposed method achieves an impressive accuracy rate of 99% in accurately detecting facial spoofing attacks.
引用
收藏
页码:935 / 942
页数:7
相关论文
共 34 条
[1]  
Velásquez I(2018)Authentication schemes and methods: a systematic literature review Inf. Softw. Technol. 94 30-37
[2]  
Caro A(2020)A review on face recognition systems: recent approaches and challenges Multimed. Tools Appl. 79 27891-27922
[3]  
Rodríguez A(2021)Countering spoof: towards detecting deepfake with multidimensional biological signals Secur Commun Netw 2021 1-8
[4]  
Oloyede MO(2014)Detecting masquerade in face recognition system—a literature survey IOSR J. Comput. Eng. 16 01-05
[5]  
Hancke GP(2000)Bioid: a multimodal biometric identification system Computer 33 64-68
[6]  
Myburgh HC(2007)Real-time face detection and motion analysis with application in liveness assessment Trans. Info. For. Sec. 2 548-558
[7]  
Jin X(2016)Joint face detection and alignment using multitask cascaded convolutional networks IEEE Signal Process. Lett. 23 1499-1503
[8]  
Ye D(2022)GANBA: generative adversarial network for biometric anti-spoofing Appl. Sci. 12 1454-1809
[9]  
Chen C(2022)Face anti-spoofing method based on residual network with channel attention mechanism Electronics 11 3056-undefined
[10]  
Lina S(2018)Unsupervised domain adaptation for face anti-spoofing IEEE Trans. Inf. Forensics Secur. 13 1794-undefined