Fig 1.
Deep FR system with face detector and alignment.
Table 1.
Different network architectures of FR.
Fig 2.
AlexNet architecture.
Fig 3.
Core knowledge of transfer learning.
Fig 4.
The general overall view of the proposed face recognition system.
Fig 5.
Block diagram of the proposed biometric system (images from dataset published in [18]).
Fig 6.
Face images before and after preprocessing (images from dataset published in [18]).
Fig 7.
The schema of the modified AlexNet, where (#S) is the number of subjects in the dataset used during training.
Fig 8.
Different types of activation functions for classification.
Fig 9.
General block diagram of the fog computing FR system.
Fig 10.
General architecture of the fog computing FR system.
Fig 11.
Fog computing network for the face recognition scheme.
Fig 12.
Face images of SDUMLA-HMT subjects under different conditions as a dataset example [18].
Table 2.
Parameter settings used in the experiments.
Fig 13.
Recognition time of the proposed FR system and individual classifiers.
Fig 14.
Precision of our proposed system and the three comparison systems.
Fig 15.
Recall of the proposed system and the three comparison systems.
Fig 16.
Accuracy of our proposed system and the three comparison systems.
Fig 17.
The specificity of the proposed system and the three comparison systems.
Table 3.
Average results of our proposed system and the three comparison systems.
Fig 18.
Average results of our proposed system and the three comparison systems.
Table 4.
Comparative accuracy details of KNN, SVM and DCNN using the SDUMLA dataset.
Fig 19.
Comparative evaluation of the proposed FR system vs recent literature.