name: "EquivNet"
input: "data"
input_dim: 1
input_dim: 3
input_dim: 227
input_dim: 227

layer{
name: "conv1"
type: "Convolution"
bottom: "data"
top: "cls_conv1"
param{
name: "conv1_w"
lr_mult: 1.0
decay_mult: 0.0
}

param{
name: "conv1_b"
lr_mult: 2.0
decay_mult: 0.0
}

convolution_param{
num_output: 96
kernel_size: 11
stride: 4
weight_filler{
type: "gaussian"
std: 0.01
}

bias_filler{
type: "constant"
value: 0
}

}

}

layer{
name: "relu1"
type: "ReLU"
bottom: "cls_conv1"
top: "cls_conv1"
}

layer{
name: "pool1"
type: "Pooling"
bottom: "cls_conv1"
top: "cls_pool1"
pooling_param{
pool: MAX
kernel_size: 3
stride: 2
}

}

layer{
name: "bn1"
type: "BatchNorm"
bottom: "cls_pool1"
top: "cls_pool1"
param{
name: "bn1_w1"
lr_mult: 0
}

param{
name: "bn1_w2"
lr_mult: 0
}

param{
name: "bn1_w3"
lr_mult: 0
}
batch_norm_param { use_global_stats: true}
}

layer{
name: "norm1"
type: "LRN"
bottom: "cls_pool1"
top: "cls_norm1"
lrn_param{
local_size: 5
alpha: 0.0001
beta: 0.75
}

}

layer{
name: "conv2"
type: "Convolution"
bottom: "cls_norm1"
top: "cls_conv2"
param{
name: "conv2_w"
lr_mult: 1.0
decay_mult: 0.0
}

param{
name: "conv2_b"
lr_mult: 2.0
decay_mult: 0.0
}

convolution_param{
num_output: 256
pad: 2
kernel_size: 5
group: 2
weight_filler{
type: "gaussian"
std: 0.01
}

bias_filler{
type: "constant"
value: 0
}

}

}

layer{
name: "relu2"
type: "ReLU"
bottom: "cls_conv2"
top: "cls_conv2"
}

layer{
name: "pool2"
type: "Pooling"
bottom: "cls_conv2"
top: "cls_pool2"
pooling_param{
pool: MAX
kernel_size: 3
stride: 2
}

}

layer{
name: "norm2"
type: "LRN"
bottom: "cls_pool2"
top: "cls_norm2"
lrn_param{
local_size: 5
alpha: 0.0001
beta: 0.75
}

}

layer{
name: "bn2"
type: "BatchNorm"
bottom: "cls_norm2"
top: "cls_norm2"
param{
name: "bn2_w1"
lr_mult: 0
}

param{
name: "bn2_w2"
lr_mult: 0
}

param{
name: "bn2_w3"
lr_mult: 0
}
batch_norm_param { use_global_stats: true}
}

layer{
name: "conv3"
type: "Convolution"
bottom: "cls_norm2"
top: "cls_conv3"
param{
name: "conv3_w"
lr_mult: 1.0
decay_mult: 0.0
}

param{
name: "conv3_b"
lr_mult: 2.0
decay_mult: 0.0
}

convolution_param{
num_output: 384
pad: 1
kernel_size: 3
weight_filler{
type: "gaussian"
std: 0.01
}

bias_filler{
type: "constant"
value: 0
}

}

}

layer{
name: "relu3"
type: "ReLU"
bottom: "cls_conv3"
top: "cls_conv3"
}

layer{
name: "bn3"
type: "BatchNorm"
bottom: "cls_conv3"
top: "cls_conv3"
param{
name: "bn3_w1"
lr_mult: 0
}

param{
name: "bn3_w2"
lr_mult: 0
}

param{
name: "bn3_w3"
lr_mult: 0
}
batch_norm_param { use_global_stats: true}
}

layer{
name: "conv4"
type: "Convolution"
bottom: "cls_conv3"
top: "cls_conv4"
param{
name: "conv4_w"
lr_mult: 1.0
decay_mult: 0.0
}

param{
name: "conv4_b"
lr_mult: 2.0
decay_mult: 0.0
}

convolution_param{
num_output: 384
pad: 1
kernel_size: 3
group: 2
weight_filler{
type: "gaussian"
std: 0.01
}

bias_filler{
type: "constant"
value: 0
}

}

}

layer{
name: "relu4"
type: "ReLU"
bottom: "cls_conv4"
top: "cls_conv4"
}

layer{
name: "bn4"
type: "BatchNorm"
bottom: "cls_conv4"
top: "cls_conv4"
param{
name: "bn4_w1"
lr_mult: 0
}

param{
name: "bn4_w2"
lr_mult: 0
}

param{
name: "bn4_w3"
lr_mult: 0
}
batch_norm_param { use_global_stats: true}
}

layer{
name: "conv5"
type: "Convolution"
bottom: "cls_conv4"
top: "cls_conv5"
param{
name: "conv5_w"
lr_mult: 1.0
decay_mult: 0.0
}

param{
name: "conv5_b"
lr_mult: 2.0
decay_mult: 0.0
}

convolution_param{
num_output: 256
pad: 1
kernel_size: 3
group: 2
weight_filler{
type: "gaussian"
std: 0.01
}

bias_filler{
type: "constant"
value: 0
}

}

}

layer{
name: "relu5"
type: "ReLU"
bottom: "cls_conv5"
top: "cls_conv5"
}

layer{
name: "pool5"
type: "Pooling"
bottom: "cls_conv5"
top: "cls_pool5"
pooling_param{
pool: MAX
kernel_size: 3
stride: 2
}

}

layer{
name: "bn5"
type: "BatchNorm"
bottom: "cls_pool5"
top: "cls_pool5"
param{
name: "bn5_w1"
lr_mult: 0
}

param{
name: "bn5_w2"
lr_mult: 0
}

param{
name: "bn5_w3"
lr_mult: 0
}
batch_norm_param { use_global_stats: true}
}

layer{
name: "drop-pool5"
type: "Power"
bottom: "cls_pool5"
top: "cls_pool5"
power_param{
scale: 0.5
}

}

layer{
name: "fc6"
type: "InnerProduct"
bottom: "cls_pool5"
top: "cls_fc6"
param{
name: "fc6_w"
lr_mult: 1
decay_mult: 1
}

param{
name: "fc6_b"
lr_mult: 2
decay_mult: 0
}

inner_product_param{
num_output: 4096
weight_filler{
type: "gaussian"
std: 0.005
}

bias_filler{
type: "constant"
value: 0
}
}
}

layer{
name: "relu6"
type: "ReLU"
bottom: "cls_fc6"
top: "cls_fc6"
}

layer{
name: "bn6"
type: "BatchNorm"
bottom: "cls_fc6"
top: "cls_fc6"
param{
name: "bn6_w1"
lr_mult: 0
}

param{
name: "bn6_w2"
lr_mult: 0
}

param{
name: "bn6_w3"
lr_mult: 0
}
batch_norm_param { use_global_stats: true}
}