Table 1 Summary of research works on the AFID dataset focuses on patch-based VIT and transfer learning models.

Cao et al.¹⁷

VIT

Unknown

Mean Square Error

Accuracy: 94.5%;

ROC AUC: 97.9%

Rabbi et al.¹⁸

VGG19,

InceptionV3,

DenseNet201

Unknown

Unknown

Accuracy: 85.0%;

ROC AUC: 92.3%,

Accuracy: 78.0%;

ROC AUC: 85.9%,

Accuracy: 83.0%;

ROC AUC: 91.0%

Alkahtani et al.¹⁹

MobileNet,

VGG-16

softmax

Cross Entropy

Accuracy: 92.0%;

Recall: 92.0%;

F1 score: 92.0%,

Accuracy: 82.1%;

Recall: 82.0%;

F1 score: 82.0%

Alam et al.¹⁶

Xception,

ResNet-50

Unknown

Cross Entropy

Accuracy: 95.0%;

ROC AUC: 98.0%;

Precision: 95.0%,

Accuracy: 94.0%;

ROC AUC: 96.0%;

Precision: 94.0%

Mujeeb Rahman and Subashini²⁰

Xception,

EfficientNetB1

softmax

Cross Entropy

Accuracy: 90.0%;

Recall: 88.4%;

Specificity: 91.6%;

ROC AUC: 96.6%,

Accuracy: 89.6%;

Recall: 86.0%;

Specificity: 94.0%;

ROC AUC: 95.0%

Alsaade and Alzahrani¹⁴

Xception,

VGG-19

softmax

Unknown

Accuracy: 91.0%;

Recall: 88.0%;

Specificity: 94.0%,

Accuracy: 80.0%;

Recall: 78.0%;

Specificity: 83.0%

Hosseini et al.¹⁵

MobileNet

softmax

Unknown

Accuracy: 94.6%

Akter et al.¹³

MobileNet,

DenseNet-121

Unknown

Unknown

Accuracy: 92.1%

Accuracy: 83.6%;

Recall: 83.6%;

Specificity: 83.6%