Layer No.

Layer Name

Briefly Explanation

1

Image Input (data)

227x227x3 images with zerocenter normalization

2

Convolution (conv1)

96 11x11x3 convolutions with stride [4 4] and padding [0 0 0 0]

3

ReLU (relu1)

ReLU

4

Cross Channel Normalization (norm1)

cross channel normalization with 5 channels per element

5

Max Pooling (pool1)

3x3 max pooling with stride [2 2] and padding [0 0 0 0]

6

Convolution (conv2)

256 5x5x48 convolutions with stride [1 1] and padding [2 2 2 2]

7

ReLU (relu2)

ReLU

8

Cross ChannelNormalization (norm2)

cross channel normalization with 5 channels per element

9

Max Pooling (pool2)

3x3 max pooling with stride [2 2] and padding [0 0 0 0]

10

Convolution (conv3)

384 3x3x256 convolutions with stride [1 1] and padding [1 1 1 1]

11

ReLU (relu3)

ReLU

12

Convolution (conv4)

384 3x3x192 convolutions with stride [1 1] and padding [1 1 1 1]

13

ReLU (relu4)

ReLU

14

Convolution (conv5)

256 3x3x192 convolutions with stride [1 1] and padding [1 1 1 1]

15

ReLU (relu5)

ReLU

16

Max Pooling (pool5)

3x3 max pooling with stride [2 2] and padding [0 0 0 0]

17

Fully Connected (fc6)

4096 fully connected layer

18

ReLU (relu6)

ReLU

19

Dropout (drop6)

50% dropout

20

Fully Connected (fc7)

4096 fully connected layer

21

ReLU (relu7)

ReLU

22

Dropout (drop7)

Dropout

23

Fully Connected (fc8)

1000 fully connected layer

24

Softmax (prob)

Softmax

25

Classification Output (output)

crossentropyex with tench and 999 other classes