This study provides a comparative evaluation of face recognition databases and Convolutional Neural Network (CNN) architectures used in training and testing face recognition systems. The databases span from early datasets like Olivetti Research Laboratory (ORL) and Facial Recognition Technology (FERET) to more recent collections such as MegaFace and Ms-Celeb-1M, offering a range of sizes, subject diversity, and image quality. Older databases, such as ORL and FERET, are smaller and cleaner, while newer datasets enable large-scale training with millions of images but pose challenges like inconsistent data quality and high computational costs. The study also examines CNN architectures, including FaceNet and Visual Geometry Group 16 (VGG16), which show strong performance on large datasets like Labeled Faces in the Wild (LFW) and VGGFace, achieving accuracy rates above 98%. In contrast, earlier models like Support Vector Machine (SVM) and Gabor Wavelets perform well on smaller datasets but lack scalability for larger, more complex datasets. The analysis highlights the growing importance of multi-task learning and ensemble methods, as seen in Multi-Task Cascaded Convolutional Networks (MTCNNs). Overall, the findings emphasize the need for advanced algorithms capable of handling large-scale, real-world challenges while optimizing accuracy and computational efficiency in face recognition systems.

Deep learning has led to the creation of facial recognition technologies using convolutional neural networks (CNNs). This preliminary study explores the application of CNN architectures in face recognition to gain a deeper understanding of the challenges and methodologies in the field. The study systematically reviewed relevant literature using the Preferred Reporting Items for Systematic reviews and Meta‐Analyses extension for Scoping Reviews (PRISMA‐ScR) framework. Out of 3622 eligible papers, 266 were included in the review, with 47% proposing new techniques and 1% focusing on method implementation and comparison. Most studies used images rather than video as training or testing data, with 78% using clean data and only 7% utilizing occluded and clean data. It was observed that traditional CNN architectures were predominantly employed. The study identified a lack of research on the implementation and definition of CNN architectures, the development of facial recognition models using both clean and occluded images and videos, and the exploration of nontraditional CNN architectures. The challenges affecting facial recognition included occlusion, distance from the camera, camera angle, and lighting conditions. This preliminary assessment provides an insight into the use of CNN in face recognition and suggests that nontraditional CNN architectures could be further explored in future research.

Face recognition systems are widely used in surveillance, biometric authentication, access control, and digital identity verification; however, supervision sensitivity, evaluation stability, and performance consistency across datasets remain insufficiently understood. This study investigates the behavior of convolutional, transformer-based, and hybrid face recognition architectures under both Softmax and ArcFace supervision using five-fold subject-disjoint cross-validation on the Labeled Faces in the Wild (LFW) and FAGEv2 datasets. ResNet50, MobileNetV3, DeiT-Small, and a Hybrid multi-branch architecture integrating complementary convolutional and transformer feature representations were evaluated using Top-1 identification accuracy, Area Under the ROC Curve (AUC), Equal Error Rate (EER), computational complexity, and fold-level statistical analysis. Experimental results revealed substantial supervision sensitivity across architectures and datasets. On the LFW dataset, Hybrid-Softmax achieved the highest Top-1 identification accuracy (62.4%), while DeiT-Small-Softmax achieved the strongest verification performance with an AUC of 0.905 and EER of 0.159. On the FAGEv2 dataset, Hybrid-Softmax and DeiT-Small-Softmax achieved the highest identification accuracy (38.0%), while Hybrid-Softmax achieved the strongest verification performance with an AUC of 0.825 and EER of 0.251. Fold-level analyses demonstrated that the effect of ArcFace supervision varied across architectures and datasets, with consistent improvements observed for some convolutional architectures but not for transformer-based or hybrid models. Cross-dataset evaluation further revealed changes in model ranking and supervision behavior, indicating that comparative performance is strongly influenced by dataset characteristics and evaluation conditions. The findings demonstrate that additive angular margin supervision does not universally outperform conventional Softmax optimization and highlight the importance of multi-dataset benchmarking, fold-level evaluation, and supervision sensitivity analysis for robust and reproducible face recognition benchmarking.

Face analysis systems are widely used in security, authentication, and public-sector applications; however, demographic bias and the statistical reliability of reported performance remain key concerns. Many studies rely on aggregate accuracy without quantifying subgroup disparities or uncertainty, potentially overstating model fairness. This study presents a statistically grounded evaluation of demographic bias in face attribute classification across three representative architectures, ResNet50, MobileNetV3, and a vision transformer (DeiT), using the FairFace and UTKFace datasets. Subgroup analysis is conducted across race and gender, incorporating disparity indices, bootstrap confidence intervals, and inferential statistical testing with effect size analysis. The evaluation uses an embedding-based nearest-neighbor approach to examine representation-level behavior consistently across models. Results show that race-based disparities are substantially larger than gender-based disparities across both datasets. On FairFace, race disparity gaps range from 0.1124 to 0.1266, while on UTKFace they increase significantly to 0.4726–0.4944, with large effect sizes (Cohen's d >1). In contrast, gender disparities remain smaller, with gaps between 0.0280 and 0.0582 on FairFace and 0.0194–0.0326 on UTKFace, and correspondingly small effect sizes ( d < 0.13). Despite modest differences in overall accuracy across models, subgroup disparities remain statistically significant across all architectures. These findings emphasize the importance of subgroup-level evaluation, uncertainty quantification, and statistical validation for reliable fairness assessment in face analysis systems.

Deep facial representation learning has achieved remarkable performance in computer vision applications; however, high overall accuracy does not necessarily ensure equitable performance across demographic groups. This study presents a unified framework for evaluating the quality and demographic fairness of pretrained deep facial representations. ResNet50, MobileNetV3, and the Data-efficient Image Transformer (DeiT) were evaluated using ImageNet-pretrained weights on the FairFace and UTKFace datasets, representing balanced and unconstrained demographic settings, respectively. Feature embeddings were assessed using one-nearest neighbour (1-NN) retrieval to measure the intrinsic discriminative capability of the representation space and Logistic Probe classification to evaluate the linear separability of demographic information. Fairness was examined through overall accuracy, subgroup performance, race–gender intersectional analysis, fairness gap, coefficient of variation, and statistical significance testing. Logistic Probe classification consistently outperformed 1-NN retrieval across both datasets, indicating that pretrained facial representations encode demographic information that can be effectively separated using simple linear classifiers. ResNet50 achieved the highest overall classification performance, whereas MobileNetV3 demonstrated more consistent subgroup performance in several evaluation settings. Intersectional analysis revealed persistent demographic disparities despite improvements in overall accuracy, underscoring the importance of fairness evaluation beyond aggregate metrics. The proposed framework provides a reproducible methodology for jointly assessing representation quality, demographic fairness, and statistical significance in deep facial representation learning.