Table 2 VGG-16 for tongue image quality control
From: Tongue image quality assessment based on a deep convolutional neural network
Layers | Feature map size | Structure |
|---|---|---|
Conv1 | 400 × 400 | 3 × 3 conv, 64 |
Conv2 | 400 × 400 | 3 × 3 conv, 64 |
Pool1 | 200 × 200 | 2 × 2 max pool, stride 2 |
Conv3 | 200 × 200 | 3 × 3 conv, 128 |
Conv4 | 200 × 200 | 3 × 3 conv, 128 |
Pool2 | 100 × 100 | 2 × 2 max pool, stride 2 |
Conv5 | 100 × 100 | 3 × 3 conv, 256 |
Conv6 | 100 × 100 | 3 × 3 conv, 256 |
Conv7 | 100 × 100 | 3 × 3 conv, 256 |
Pool3 | 50 × 50 | 2 × 2 max pool, stride 2 |
Conv8 | 50 × 50 | 3 × 3 conv, 512 |
Conv9 | 50 × 50 | 3 × 3 conv, 512 |
Conv10 | 50 × 50 | 3 × 3 conv, 512 |
Pool4 | 25 × 25 | 2 × 2 max pool, stride 2 |
Conv11 | 25 × 25 | 3 × 3 conv, 512 |
Conv12 | 25 × 25 | 3 × 3 conv, 512 |
Conv13 | 25 × 25 | 3 × 3 conv, 512 |
Pool5 | 12 × 12 | 2 × 2 max pool, stride 2 |
Classification Layer | 1 × 1 | 2208D fully connected layer with ReLU |
2D fully connected layer | ||
Softmax |