Table 1 Details of DenseNet-121 CNN Architecture.
From: Uncertainty-aware diabetic retinopathy detection using deep learning enhanced by Bayesian approaches
Layers | Output size | DenseNet-121 |
|---|---|---|
Convolution | 112x112 | 7x7 Conv, stride 2 |
Pooling | 56x56 | 3x3 Max Pooling, stride 2 |
Dense Block (1) | 56x56 | [1x1 conv, 3x3 conv] x 6 |
Transition Layer (1) | 56x56 | [1x1 conv], 2x2 Avg Pool, stride 2 |
Dense Block (2) | 28x28 | [1x1 conv, 3x3 conv] x 12 |
Transition Layer (2) | 28x28 | [1x1 conv], 2x2 Avg Pool, stride 2 |
Dense Block (3) | 14x14 | [1x1 conv, 3x3 conv] x 24 |
Transition Layer (3) | 14x14 | [1x1 conv], 2x2 Avg Pool, stride 2 |
Dense Block (4) | 7x7 | [1x1 conv, 3x3 conv] x 16 |
Classification Layers | 1x1 | 7x7 Global Avg Pool, FC Layer, Bayesian method Layers, Softmax |
