Table 2 Detailed descriptions of AlexNet, VGG-16, and GoogLeNet
From: Foreign object debris material recognition based on convolutional neural networks
| Â | Alexnet | VGG16 | GoogLeNet | Improved AlexNet |
|---|---|---|---|---|
Input (RGB image) | 227*227 | 224*224 | 224*224 | 227*227 |
Convolution (kernel size/stride) | 11*11/4 | 3*3/1 | 7*7/2 | 11*11/4 |
3*3/1 | ||||
Max. pooling | 3*3/2 | 2*2/2 | 3*3/2 | 3*3/2 |
Convolution (kernel size/stride) | 5*5/1 | 3*3/1 | 3*3/1 | 5*5/1 |
3*3/1 | ||||
Max. pooling | 3*3/2 | 2*2/2 | 3*3/2 | 3*3/2 |
Convolution (kernel size/stride) | 3*3/1 | 3*3/1 | Inception(3a) | 3*3/1 |
3*3/1 | 3*3/1 | Inception(3b) | 3*3/1 | |
3*3/1 | 1*1/1 | 3*3/1 | ||
Max. pooling | 3*3/2 | 2*2/2 | 3*3/2 | 3*3/2 |
Convolution (kernel size/stride) | Â | 3*3/1 | Inception(4a) | Â |
3*3/1 | Inception(4b) | |||
1*1/1 | Inception(4c) | |||
Inception(4d) | ||||
Inception(4e) | ||||
Max. pooling | Â | 2*2/2 | 3*3/2 | Â |
Convolution (kernel size/stride) | Â | 3*3/1 | Inception(5a) | Â |
3*3/1 | Inception(5b) | |||
1*1/1 | ||||
Pooling | Â | Max. pool 2*2/2 | Average pool 7*7/1 | Â |
Linear | FC-4096 | FC-4096 | FC-1000 | FC-4096 |
FC-4096 | FC-4096 | FC-4096 | ||
FC-1000 | FC-1000 | FC-1000 | ||
FC-3 | ||||
Output | Softmax | Softmax | Softmax | Softmax |