Layer Name

Type

Description

“input”

mage Input

224 × 224 × 3 images with “zerocenter” normalization

“conv1_1”

Convolution

64 3 × 3 × 3 convolutions with stride [1 1] and padding [1 1]

“conv1_2”

Convolution

64 3 × 3 × 64 convolutions with stride [1 1] and padding [1 1]

“pool1”

Max Pooling

2 × 2 max pooling with stride [2 2] and padding [0 0]

“conv2_1”

Convolution

128 3 × 3 × 64 convolutions with stride [1 1] and padding [1 1]

“conv2_2”

Convolution

128 3 × 3 × 128 convolutions with stride [1 1] and padding [1 1]

“pool2”

Max Pooling

2 × 2 max pooling with stride [2 2] and padding [0 0]

“conv3_1”

Convolution

256 3 × 3 × 128 convolutions with stride [1 1] and padding [1 1]

“conv3_2”

Convolution

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

“conv3_3”

Convolution

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

“pool3”

Max Pooling

2 × 2 max pooling with stride [2 2] and padding [0 0]

“conv4_1”

Convolution

512 3 × 3 × 256 convolutions with stride [1 1] and padding [1 1]

“conv4_2”

Convolution

512 3 × 3 × 512 convolutions with stride [1 1] and padding [1 1]

“conv4_3”

Convolution

512 3 × 3 × 512 convolutions with stride [1 1] and padding [1 1]

“pool4”

Max Pooling

2 × 2 max pooling with stride [2 2] and padding [0 0]

“conv5_1”

Convolution

512 3 × 3 × 512 convolutions with stride [1 1] and padding [1 1]

“conv5_2”

Convolution

512 3 × 3 × 512 convolutions with stride [1 1] and padding [1 1]

“conv5_3”

Convolution

512 3 × 3 × 512 convolutions with stride [1 1] and padding [1 1]

“pool5”

Max Pooling

2x2 max pooling with stride [2 2] and padding [0 0]

“fc6”

Fully Connected

4096 fully connected layer

“drop6”

Dropout

50% dropout

“fc7”

Fully Connected

4096 fully connected layer

“drop7”

Dropout

50% dropout

“fc8”

Fully Connected

1000 fully connected layer

“prob”

Softmax

softmax

“output”

Classification Output

crossentropyex with “tench”, “goldfish”, and 998 other classes