Table 3 The structure of VGG16-Net
From: Precipitation cloud identification based on faster-RCNN for Doppler weather radar
Layer | Kernal | Input | Output |
|---|---|---|---|
Conv1(Conv2D) | (1,3,3,64) | (1,400, 400, 5/7) | (1,400,400,64) |
Conv2(Conv2D) | (1,3,3,64) | (1,400, 400, 64) | (1,400,400,64) |
maxPooling1 | (2,2) | (1,400, 400, 64) | (1,200, 200, 64) |
Conv3(Conv2D) | (1,3,3,128) | (1,200, 200, 64) | (1,200, 200,128) |
Conv4(Conv2D) | (1,3,3,128) | (1,200,200,128) | (1,200, 200,128) |
maxPooling2 | (2,2) | (1,200,200,128) | (1,100,100,128) |
Conv5(Conv2D) | (1,3,3,256) | (1,100,100, 128) | (1,100,100,256) |
Conv6(Conv2D) | (1,3,3,256) | (1,100,100, 256) | (1,100,100,256) |
Conv7(Conv2D) | (1,3,3,256) | (1,100,100, 256) | (1,100,100,256) |
maxPooling3 | (2,2) | (1,100,100, 256) | (1,50, 50, 256) |
Conv8(Conv2D) | (1,3,3,512) | (1,50, 50, 256) | (1,50, 50,512) |
Conv9(Conv2D) | (1,3,3,512) | (1,50, 50,512) | (1,50, 50,512) |
maxPooling4 | (2,2) | (1,50, 50,512) | (1,25,25,512) |
Conv11(Conv2D) | (1,3,3,512) | (1,25,25,512) | (1,25,25,512) |
Conv12(Conv2D) | (1,3,3,512) | (1,25,25,512) | (1,25,25,512) |
Conv13(Conv2D) | (1,3,3,512) | (1,25,25,512) | (1,25,25,512) |