Table 5 Comparison of suggested models with existing DR classification research.
From: Uncertainty-aware diabetic retinopathy detection using deep learning enhanced by Bayesian approaches
Reference | Year | Method | Accuracy |
|---|---|---|---|
Filos et al.36 | 2019 | Ensemble MC-Dropout MFVI | 87.1% 81.1% |
Nguyen et al.37 | 2020 | VGG-16 VGG-19 | 80% 82% |
Taufiqurrahman et al.38 | 2020 | MobileNetV2-SVM | 85% |
Yi et al.39 | 2021 | Residual Attention EfficientNet | 93.5% |
Gangwar & Ravi40 | 2021 | Hybrid Inception-ResNet-v2 | 82.18% |
Islam et al.41 | 2022 | SCL (Supervised Contrastive Learning) | 84% |
Alahmadi42 | 2022 | Recalibration of style and content by DL | 85% |
Oulhadj et al.43 | 2022 | DenseNet, InceptionV3, ResNet-50 | 85.28% |
Butt et al.44 | 2022 | ResNet-18 GoogleNet | 89.29% |
A.M. Fayyaz et al.45 | 2023 | AlexNet ResNet-101 | 93.0% |
Tiwari46 | 2023 | ResNet50 | 91.60% |
W. K. Wong47 | 2023 | ShuffleNet ResNet-18 | 82% 75% |
A. Jabbar48 | 2024 | GoogleNet + ResNet | 94% |
Current Work | – | BCNN-MC Dropout BCNN-MFVI BCNN-Deterministic | 97.68% 94.23% 91.44% |
