Face Presentation Attack Detection Using Deep Background Subtraction

Benlamoudi, Azeddine; Bekhouche, Salah Eddine; Korichi, Maarouf; Bensid, Khaled; Ouahabi, Abdeldjalil; Hadid, Abdenour; Taleb‐Ahmed, Abdelmalik

doi:10.3390/s22103760

Cited by 14 publications

(6 citation statements)

References 55 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The different results, and the comparison with state-of-the-art methods, show that our approach can be a useful tool for brain cancer detection, diagnosis, and radiotherapy treatment planning. The future direction of our research in brain tumor segmentation must address the limitations of the unsupervised approach by: (1) combining PSO, ANOVA, and a CNN model [ 84 , 85 , 86 , 87 , 88 , 89 , 90 ]; (2) using generative adversarial networks [ 91 , 92 , 93 , 94 ] to pre-process, colorize, correct, and enhance images before presenting them to the segmentation algorithm.…”

Section: Discussionmentioning

confidence: 99%

Particle Swarm Optimization and Two-Way Fixed-Effects Analysis of Variance for Efficient Brain Tumor Segmentation

Atia

Benzaoui

Jacques

et al. 2022

Cancers

Self Cite

View full text Add to dashboard Cite

Segmentation of brain tumor images, to refine the detection and understanding of abnormal masses in the brain, is an important research topic in medical imaging. This paper proposes a new segmentation method, consisting of three main steps, to detect brain lesions using magnetic resonance imaging (MRI). In the first step, the parts of the image delineating the skull bone are removed, to exclude insignificant data. In the second step, which is the main contribution of this study, the particle swarm optimization (PSO) technique is applied, to detect the block that contains the brain lesions. The fitness function, used to determine the best block among all candidate blocks, is based on a two-way fixed-effects analysis of variance (ANOVA). In the last step of the algorithm, the K-means segmentation method is used in the lesion block, to classify it as a tumor or not. A thorough evaluation of the proposed algorithm was performed, using: (1) a private MRI database provided by the Kouba imaging center—Algiers (KICA); (2) the multimodal brain tumor segmentation challenge (BraTS) 2015 database. Estimates of the selected fitness function were first compared to those based on the sum-of-absolute-differences (SAD) dissimilarity criterion, to demonstrate the efficiency and robustness of the ANOVA. The performance of the optimized brain tumor segmentation algorithm was then compared to the results of several state-of-the-art techniques. The results obtained, by using the Dice coefficient, Jaccard distance, correlation coefficient, and root mean square error (RMSE) measurements, demonstrated the superiority of the proposed optimized segmentation algorithm over equivalent techniques.

show abstract

Section: Discussionmentioning

confidence: 99%

Particle Swarm Optimization and Two-Way Fixed-Effects Analysis of Variance for Efficient Brain Tumor Segmentation

Atia

Benzaoui

Jacques

et al. 2022

Cancers

Self Cite

View full text Add to dashboard Cite

show abstract

“…− Using generative adversarial networks [62−65] to preprocess, colorize, correct, and enhance images before presenting them to the segmentation algorithm. − Combining the features extracted by the proposed approach with deeplearned features [66][67][68][69][70][71][72]7] to improve the quality of the segmentation and make it more semantic.…”

Section: Discussionmentioning

confidence: 99%

Particle Swarm Optimization and Two-Way Fixed-Effects Analysis of Variance for Efficient Brain Tumor Segmentation

Atia¹,

Benzaoui²,

Jacques³

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

Segmentation of brain tumor images is a major research topic in medical imaging to have a refined detection and understanding of abnormal masses in the brain. This paper proposes a new segmentation method, consisting of three main steps, to detect brain lesions using magnetic resonance imaging (MRI). In the first step, the parts of the image delineating the skull bone are removed to exclude insignificant data. In the second step, which is the main contribution of this study, the particle swarm optimization (PSO) technique is applied to detect the block that contains the brain lesions. The fitness function, used to determine the best block among all candidate blocks, is based on a two-way fixed-effects analysis of variance (ANOVA). In the last step of the algorithm, the K-means segmentation method is used in the lesion block to classify it as tumor or not. A thorough evaluation of the proposed algorithm is performed using the MRI database provided by the Kouba imaging center in Algiers, Algeria. Estimates of the selected fitness function are first compared to those based on the sum-of-absolute-differences (SAD) dissimilarity criterion and demonstrate the efficiency and robustness of the ANOVA. The performance of the optimized brain tumor segmentation algorithm is then compared to the results of several state-of-the-art techniques, including fuzzy C-means, K-means, Otsu thresholding, local thresholding, and watershed segmentation. The results obtained using Dice coefficient, Jaccard distance, correlation coefficient, and root mean square error (RMSE) measurements demonstrate the superiority of the proposed optimized segmentation algorithm over equivalent techniques.

show abstract

“…Regrettably, many of these methods rely on computationally expensive models, rendering them unsuitable for real-time Face Anti-Spoofing (FAS applications. In [33], Benlamoudi et al presented an approach based on background subtraction and used pre-trained ResNet-50 [34] CNN architecture to learn features related to genuine and spoof faces. Their approach is based on the assumption that the capturing camera is always static.…”

Section: B Deep Learning-based Methodsmentioning

confidence: 99%

Securing Phygital Gameplay: Strategies for Video-Replay Spoofing Detection

Huszár,

Adhikarla

2024

IEEE Access

View full text Add to dashboard Cite

Physical Virtual Sports (PVS) utilize digital technologies for the analysis and evaluation of sports performances. This research article addresses the challenge of detecting video-replay spoofing in PVS, with a specific focus on a digital football sport aimed at assessing and improving a player's football juggling skills. In the context of the growing presence of digital coaches as well as PVS, accurate assessment of player performance and identification of deceptive practices in these applications are paramount. The proliferation of sophisticated technologies, such as deepfake algorithms and computer vision techniques, has facilitated the manipulation of video replays, deceiving both viewers and officials. To tackle the challenges associated with video-replay spoofing, this article introduces a meticulously curated dataset comprising 600 players engaged in the digital football sport. Additionally, the dataset includes video-replay spoofing videos captured on a wide range of display devices. A deep learning-based model is developed and trained on this dataset, achieving an accuracy rate of approximately 95%. Generalization studies were also conducted to assess the model's ability to generalize to unseen scenarios and datasets. The ROC-AUC score highlighted the model's discriminative power across different threshold values, validating its effectiveness in distinguishing between genuine and spoofed video replays. The results demonstrate that our trained model exhibited consistent performance across multiple public face biometric spoofing datasets, underscoring its robustness against sophisticated video-replay attacks in various domains. Additionally, ablation studies were carried out by systematically removing or modifying the model's backbone architectures to analyze their effects on detection accuracy and reliability. Furthermore, computational complexity analysis was presented to evaluate the model's efficiency in terms of time and space requirements. The findings underscore the scientific significance and relevance of video replay spoof detection in PVS. By presenting a novel dataset a and employing an advanced deep learning approach, this article contributes to the scientific community's understanding and progress in combating fraudulent practices, ultimately preserving the integrity and fairness of digital sports applications.

show abstract

Face Presentation Attack Detection Using Deep Background Subtraction

Cited by 14 publications

References 55 publications

Particle Swarm Optimization and Two-Way Fixed-Effects Analysis of Variance for Efficient Brain Tumor Segmentation

Particle Swarm Optimization and Two-Way Fixed-Effects Analysis of Variance for Efficient Brain Tumor Segmentation

Particle Swarm Optimization and Two-Way Fixed-Effects Analysis of Variance for Efficient Brain Tumor Segmentation

Securing Phygital Gameplay: Strategies for Video-Replay Spoofing Detection

Contact Info

Product

Resources

About