train.prototxt

name: "resnet-blstm-ctc-ocr"
layer {
  name: "data"
  type: "Data"
  top: "data"
  top: "label"
  include {
    phase: TRAIN
  }
  data_param {
    source: "train_lmdb"
    batch_size: 16
    backend: LMDB
  }
}
layer {
  name: "data"
  type: "Data"
  top: "data"
  top: "label"
  include {
    phase: TEST
  }
  data_param {
    source: "val_lmdb"
    batch_size: 1
    backend: LMDB
  }
}

layer {
  include {
    phase: TRAIN
  }
    name: "indicator"
    type: "ContinuationIndicator"
    top: "indicator"
    continuation_indicator_param {
        time_step: 19
        batch_size: 16
    }
}
layer {
  include {
    phase: TEST
  }
    name: "indicator"
    type: "ContinuationIndicator"
    top: "indicator"
    continuation_indicator_param {
        time_step:  19
        batch_size: 1
    }
}
layer {
  name: "data_bn"
  type: "BatchNorm"
  bottom: "data"
  top: "data_bn"
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
}
layer {
  name: "data_scale"
  type: "Scale"
  bottom: "data_bn"
  top: "data_bn"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  param {
    lr_mult: 2.0
    decay_mult: 1.0
  }
  scale_param {
    bias_term: true
  }
}
layer {
  name: "conv1"
  type: "Convolution"
  bottom: "data_bn"
  top: "conv1"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  param {
    lr_mult: 2.0
    decay_mult: 1.0
  }
  convolution_param {
    num_output: 64
    pad: 3
    kernel_size: 7
    stride: 2
    weight_filler {
      type: "msra"
      variance_norm: FAN_OUT
    }
    bias_filler {
      type: "constant"
      value: 0.0
    }
  }
}
layer {
  name: "conv1_bn"
  type: "BatchNorm"
  bottom: "conv1"
  top: "conv1"
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
}
layer {
  name: "conv1_scale"
  type: "Scale"
  bottom: "conv1"
  top: "conv1"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  param {
    lr_mult: 2.0
    decay_mult: 1.0
  }
  scale_param {
    bias_term: true
  }
}
layer {
  name: "conv1_relu"
  type: "ReLU"
  bottom: "conv1"
  top: "conv1"
}
layer {
  name: "conv1_pool"
  type: "Pooling"
  bottom: "conv1"
  top: "conv1_pool"
  pooling_param {
    kernel_size: 3
    stride: 2
  }
}
layer {
  name: "layer_64_1_conv1"
  type: "Convolution"
  bottom: "conv1_pool"
  top: "layer_64_1_conv1"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  convolution_param {
    num_output: 64
    bias_term: false
    pad: 0
    kernel_size: 1
    stride: 1
    weight_filler {
      type: "msra"
    }
    bias_filler {
      type: "constant"
      value: 0.0
    }
  }
}
layer {
  name: "layer_64_1_bn2"
  type: "BatchNorm"
  bottom: "layer_64_1_conv1"
  top: "layer_64_1_conv1"
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
}
layer {
  name: "layer_64_1_scale2"
  type: "Scale"
  bottom: "layer_64_1_conv1"
  top: "layer_64_1_conv1"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  param {
    lr_mult: 2.0
    decay_mult: 1.0
  }
  scale_param {
    bias_term: true
  }
}
layer {
  name: "layer_64_1_relu2"
  type: "ReLU"
  bottom: "layer_64_1_conv1"
  top: "layer_64_1_conv1"
}
layer {
  name: "layer_64_1_conv2"
  type: "Convolution"
  bottom: "layer_64_1_conv1"
  top: "layer_64_1_conv2"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  convolution_param {
    num_output: 64
    bias_term: false
    pad: 1
    kernel_size: 3
    stride: 1
    weight_filler {
      type: "msra"
    }
    bias_filler {
      type: "constant"
      value: 0.0
    }
  }
}
layer {
  name: "layer_64_1_bn3"
  type: "BatchNorm"
  bottom: "layer_64_1_conv2"
  top: "layer_64_1_conv2"
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
}
layer {
  name: "layer_64_1_scale3"
  type: "Scale"
  bottom: "layer_64_1_conv2"
  top: "layer_64_1_conv2"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  param {
    lr_mult: 2.0
    decay_mult: 1.0
  }
  scale_param {
    bias_term: true
  }
}
layer {
  name: "layer_64_1_relu3"
  type: "ReLU"
  bottom: "layer_64_1_conv2"
  top: "layer_64_1_conv2"
}
layer {
  name: "layer_64_1_conv3"
  type: "Convolution"
  bottom: "layer_64_1_conv2"
  top: "layer_64_1_conv3"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  convolution_param {
    num_output: 256
    bias_term: false
    pad: 0
    kernel_size: 1
    stride: 1
    weight_filler {
      type: "msra"
    }
    bias_filler {
      type: "constant"
      value: 0.0
    }
  }
}
layer {
  name: "layer_64_1_conv_expand"
  type: "Convolution"
  bottom: "layer_64_1_conv1"
  top: "layer_64_1_conv_expand"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  convolution_param {
    num_output: 256
    bias_term: false
    pad: 0
    kernel_size: 1
    stride: 1
    weight_filler {
      type: "msra"
    }
    bias_filler {
      type: "constant"
      value: 0.0
    }
  }
}
layer {
  name: "layer_64_1_sum"
  type: "Eltwise"
  bottom: "layer_64_1_conv3"
  bottom: "layer_64_1_conv_expand"
  top: "layer_64_1_sum"
}
layer {
  name: "layer_64_2_bn1"
  type: "BatchNorm"
  bottom: "layer_64_1_sum"
  top: "layer_64_2_bn1"
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
}
layer {
  name: "layer_64_2_scale1"
  type: "Scale"
  bottom: "layer_64_2_bn1"
  top: "layer_64_2_bn1"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  param {
    lr_mult: 2.0
    decay_mult: 1.0
  }
  scale_param {
    bias_term: true
  }
}
layer {
  name: "layer_64_2_relu1"
  type: "ReLU"
  bottom: "layer_64_2_bn1"
  top: "layer_64_2_bn1"
}
layer {
  name: "layer_64_2_conv1"
  type: "Convolution"
  bottom: "layer_64_2_bn1"
  top: "layer_64_2_conv1"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  convolution_param {
    num_output: 64
    bias_term: false
    pad: 0
    kernel_size: 1
    stride: 1
    weight_filler {
      type: "msra"
    }
    bias_filler {
      type: "constant"
      value: 0.0
    }
  }
}
layer {
  name: "layer_64_2_bn2"
  type: "BatchNorm"
  bottom: "layer_64_2_conv1"
  top: "layer_64_2_conv1"
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
}
layer {
  name: "layer_64_2_scale2"
  type: "Scale"
  bottom: "layer_64_2_conv1"
  top: "layer_64_2_conv1"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  param {
    lr_mult: 2.0
    decay_mult: 1.0
  }
  scale_param {
    bias_term: true
  }
}
layer {
  name: "layer_64_2_relu2"
  type: "ReLU"
  bottom: "layer_64_2_conv1"
  top: "layer_64_2_conv1"
}
layer {
  name: "layer_64_2_conv2"
  type: "Convolution"
  bottom: "layer_64_2_conv1"
  top: "layer_64_2_conv2"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  convolution_param {
    num_output: 64
    bias_term: false
    pad: 1
    kernel_size: 3
    stride: 1
    weight_filler {
      type: "msra"
    }
    bias_filler {
      type: "constant"
      value: 0.0
    }
  }
}
layer {
  name: "layer_64_2_bn3"
  type: "BatchNorm"
  bottom: "layer_64_2_conv2"
  top: "layer_64_2_conv2"
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
}
layer {
  name: "layer_64_2_scale3"
  type: "Scale"
  bottom: "layer_64_2_conv2"
  top: "layer_64_2_conv2"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  param {
    lr_mult: 2.0
    decay_mult: 1.0
  }
  scale_param {
    bias_term: true
  }
}
layer {
  name: "layer_64_2_relu3"
  type: "ReLU"
  bottom: "layer_64_2_conv2"
  top: "layer_64_2_conv2"
}
layer {
  name: "layer_64_2_conv3"
  type: "Convolution"
  bottom: "layer_64_2_conv2"
  top: "layer_64_2_conv3"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  convolution_param {
    num_output: 256
    bias_term: false
    pad: 0
    kernel_size: 1
    stride: 1
    weight_filler {
      type: "msra"
    }
    bias_filler {
      type: "constant"
      value: 0.0
    }
  }
}
layer {
  name: "layer_64_2_sum"
  type: "Eltwise"
  bottom: "layer_64_2_conv3"
  bottom: "layer_64_1_sum"
  top: "layer_64_2_sum"
}
layer {
  name: "layer_64_3_bn1"
  type: "BatchNorm"
  bottom: "layer_64_2_sum"
  top: "layer_64_3_bn1"
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
}
layer {
  name: "layer_64_3_scale1"
  type: "Scale"
  bottom: "layer_64_3_bn1"
  top: "layer_64_3_bn1"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  param {
    lr_mult: 2.0
    decay_mult: 1.0
  }
  scale_param {
    bias_term: true
  }
}
layer {
  name: "layer_64_3_relu1"
  type: "ReLU"
  bottom: "layer_64_3_bn1"
  top: "layer_64_3_bn1"
}
layer {
  name: "layer_64_3_conv1"
  type: "Convolution"
  bottom: "layer_64_3_bn1"
  top: "layer_64_3_conv1"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  convolution_param {
    num_output: 64
    bias_term: false
    pad: 0
    kernel_size: 1
    stride: 1
    weight_filler {
      type: "msra"
    }
    bias_filler {
      type: "constant"
      value: 0.0
    }
  }
}
layer {
  name: "layer_64_3_bn2"
  type: "BatchNorm"
  bottom: "layer_64_3_conv1"
  top: "layer_64_3_conv1"
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
}
layer {
  name: "layer_64_3_scale2"
  type: "Scale"
  bottom: "layer_64_3_conv1"
  top: "layer_64_3_conv1"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  param {
    lr_mult: 2.0
    decay_mult: 1.0
  }
  scale_param {
    bias_term: true
  }
}
layer {
  name: "layer_64_3_relu2"
  type: "ReLU"
  bottom: "layer_64_3_conv1"
  top: "layer_64_3_conv1"
}
layer {
  name: "layer_64_3_conv2"
  type: "Convolution"
  bottom: "layer_64_3_conv1"
  top: "layer_64_3_conv2"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  convolution_param {
    num_output: 64
    bias_term: false
    pad: 1
    kernel_size: 3
    stride: 1
    weight_filler {
      type: "msra"
    }
    bias_filler {
      type: "constant"
      value: 0.0
    }
  }
}
layer {
  name: "layer_64_3_bn3"
  type: "BatchNorm"
  bottom: "layer_64_3_conv2"
  top: "layer_64_3_conv2"
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
}
layer {
  name: "layer_64_3_scale3"
  type: "Scale"
  bottom: "layer_64_3_conv2"
  top: "layer_64_3_conv2"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  param {
    lr_mult: 2.0
    decay_mult: 1.0
  }
  scale_param {
    bias_term: true
  }
}
layer {
  name: "layer_64_3_relu3"
  type: "ReLU"
  bottom: "layer_64_3_conv2"
  top: "layer_64_3_conv2"
}
layer {
  name: "layer_64_3_conv3"
  type: "Convolution"
  bottom: "layer_64_3_conv2"
  top: "layer_64_3_conv3"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  convolution_param {
    num_output: 256
    bias_term: false
    pad: 0
    kernel_size: 1
    stride: 1
    weight_filler {
      type: "msra"
    }
    bias_filler {
      type: "constant"
      value: 0.0
    }
  }
}
layer {
  name: "layer_64_3_sum"
  type: "Eltwise"
  bottom: "layer_64_3_conv3"
  bottom: "layer_64_2_sum"
  top: "layer_64_3_sum"
}
layer {
  name: "layer_128_1_bn1"
  type: "BatchNorm"
  bottom: "layer_64_3_sum"
  top: "layer_128_1_bn1"
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
}
layer {
  name: "layer_128_1_scale1"
  type: "Scale"
  bottom: "layer_128_1_bn1"
  top: "layer_128_1_bn1"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  param {
    lr_mult: 2.0
    decay_mult: 1.0
  }
  scale_param {
    bias_term: true
  }
}
layer {
  name: "layer_128_1_relu1"
  type: "ReLU"
  bottom: "layer_128_1_bn1"
  top: "layer_128_1_bn1"
}
layer {
  name: "layer_128_1_conv1"
  type: "Convolution"
  bottom: "layer_128_1_bn1"
  top: "layer_128_1_conv1"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  convolution_param {
    num_output: 128
    bias_term: false
    pad: 0
    kernel_size: 1
    stride: 1
    weight_filler {
      type: "msra"
    }
    bias_filler {
      type: "constant"
      value: 0.0
    }
  }
}
layer {
  name: "layer_128_1_bn2"
  type: "BatchNorm"
  bottom: "layer_128_1_conv1"
  top: "layer_128_1_conv1"
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
}
layer {
  name: "layer_128_1_scale2"
  type: "Scale"
  bottom: "layer_128_1_conv1"
  top: "layer_128_1_conv1"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  param {
    lr_mult: 2.0
    decay_mult: 1.0
  }
  scale_param {
    bias_term: true
  }
}
layer {
  name: "layer_128_1_relu2"
  type: "ReLU"
  bottom: "layer_128_1_conv1"
  top: "layer_128_1_conv1"
}
layer {
  name: "layer_128_1_conv2"
  type: "Convolution"
  bottom: "layer_128_1_conv1"
  top: "layer_128_1_conv2"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  convolution_param {
    num_output: 128
    bias_term: false
    pad: 1
    kernel_size: 3
    stride: 2
    weight_filler {
      type: "msra"
    }
    bias_filler {
      type: "constant"
      value: 0.0
    }
  }
}
layer {
  name: "layer_128_1_bn3"
  type: "BatchNorm"
  bottom: "layer_128_1_conv2"
  top: "layer_128_1_conv2"
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
}
layer {
  name: "layer_128_1_scale3"
  type: "Scale"
  bottom: "layer_128_1_conv2"
  top: "layer_128_1_conv2"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  param {
    lr_mult: 2.0
    decay_mult: 1.0
  }
  scale_param {
    bias_term: true
  }
}
layer {
  name: "layer_128_1_relu3"
  type: "ReLU"
  bottom: "layer_128_1_conv2"
  top: "layer_128_1_conv2"
}
layer {
  name: "layer_128_1_conv3"
  type: "Convolution"
  bottom: "layer_128_1_conv2"
  top: "layer_128_1_conv3"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  convolution_param {
    num_output: 512
    bias_term: false
    pad: 0
    kernel_size: 1
    stride: 1
    weight_filler {
      type: "msra"
    }
    bias_filler {
      type: "constant"
      value: 0.0
    }
  }
}
layer {
  name: "layer_128_1_conv_expand"
  type: "Convolution"
  bottom: "layer_128_1_bn1"
  top: "layer_128_1_conv_expand"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  convolution_param {
    num_output: 512
    bias_term: false
    pad: 0
    kernel_size: 1
    stride: 2
    weight_filler {
      type: "msra"
    }
    bias_filler {
      type: "constant"
      value: 0.0
    }
  }
}
layer {
  name: "layer_128_1_sum"
  type: "Eltwise"
  bottom: "layer_128_1_conv3"
  bottom: "layer_128_1_conv_expand"
  top: "layer_128_1_sum"
}
layer {
  name: "layer_128_2_bn1"
  type: "BatchNorm"
  bottom: "layer_128_1_sum"
  top: "layer_128_2_bn1"
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
}
layer {
  name: "layer_128_2_scale1"
  type: "Scale"
  bottom: "layer_128_2_bn1"
  top: "layer_128_2_bn1"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  param {
    lr_mult: 2.0
    decay_mult: 1.0
  }
  scale_param {
    bias_term: true
  }
}
layer {
  name: "layer_128_2_relu1"
  type: "ReLU"
  bottom: "layer_128_2_bn1"
  top: "layer_128_2_bn1"
}
layer {
  name: "layer_128_2_conv1"
  type: "Convolution"
  bottom: "layer_128_2_bn1"
  top: "layer_128_2_conv1"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  convolution_param {
    num_output: 128
    bias_term: false
    pad: 0
    kernel_size: 1
    stride: 1
    weight_filler {
      type: "msra"
    }
    bias_filler {
      type: "constant"
      value: 0.0
    }
  }
}
layer {
  name: "layer_128_2_bn2"
  type: "BatchNorm"
  bottom: "layer_128_2_conv1"
  top: "layer_128_2_conv1"
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
}
layer {
  name: "layer_128_2_scale2"
  type: "Scale"
  bottom: "layer_128_2_conv1"
  top: "layer_128_2_conv1"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  param {
    lr_mult: 2.0
    decay_mult: 1.0
  }
  scale_param {
    bias_term: true
  }
}
layer {
  name: "layer_128_2_relu2"
  type: "ReLU"
  bottom: "layer_128_2_conv1"
  top: "layer_128_2_conv1"
}
layer {
  name: "layer_128_2_conv2"
  type: "Convolution"
  bottom: "layer_128_2_conv1"
  top: "layer_128_2_conv2"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  convolution_param {
    num_output: 128
    bias_term: false
    pad: 1
    kernel_size: 3
    stride: 1
    weight_filler {
      type: "msra"
    }
    bias_filler {
      type: "constant"
      value: 0.0
    }
  }
}
layer {
  name: "layer_128_2_bn3"
  type: "BatchNorm"
  bottom: "layer_128_2_conv2"
  top: "layer_128_2_conv2"
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
}
layer {
  name: "layer_128_2_scale3"
  type: "Scale"
  bottom: "layer_128_2_conv2"
  top: "layer_128_2_conv2"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  param {
    lr_mult: 2.0
    decay_mult: 1.0
  }
  scale_param {
    bias_term: true
  }
}
layer {
  name: "layer_128_2_relu3"
  type: "ReLU"
  bottom: "layer_128_2_conv2"
  top: "layer_128_2_conv2"
}
layer {
  name: "layer_128_2_conv3"
  type: "Convolution"
  bottom: "layer_128_2_conv2"
  top: "layer_128_2_conv3"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  convolution_param {
    num_output: 512
    bias_term: false
    pad: 0
    kernel_size: 1
    stride: 1
    weight_filler {
      type: "msra"
    }
    bias_filler {
      type: "constant"
      value: 0.0
    }
  }
}
layer {
  name: "layer_128_2_sum"
  type: "Eltwise"
  bottom: "layer_128_2_conv3"
  bottom: "layer_128_1_sum"
  top: "layer_128_2_sum"
}
layer {
  name: "layer_128_3_bn1"
  type: "BatchNorm"
  bottom: "layer_128_2_sum"
  top: "layer_128_3_bn1"
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
}
layer {
  name: "layer_128_3_scale1"
  type: "Scale"
  bottom: "layer_128_3_bn1"
  top: "layer_128_3_bn1"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  param {
    lr_mult: 2.0
    decay_mult: 1.0
  }
  scale_param {
    bias_term: true
  }
}
layer {
  name: "layer_128_3_relu1"
  type: "ReLU"
  bottom: "layer_128_3_bn1"
  top: "layer_128_3_bn1"
}
layer {
  name: "layer_128_3_conv1"
  type: "Convolution"
  bottom: "layer_128_3_bn1"
  top: "layer_128_3_conv1"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  convolution_param {
    num_output: 128
    bias_term: false
    pad: 0
    kernel_size: 1
    stride: 1
    weight_filler {
      type: "msra"
    }
    bias_filler {
      type: "constant"
      value: 0.0
    }
  }
}
layer {
  name: "layer_128_3_bn2"
  type: "BatchNorm"
  bottom: "layer_128_3_conv1"
  top: "layer_128_3_conv1"
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
}
layer {
  name: "layer_128_3_scale2"
  type: "Scale"
  bottom: "layer_128_3_conv1"
  top: "layer_128_3_conv1"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  param {
    lr_mult: 2.0
    decay_mult: 1.0
  }
  scale_param {
    bias_term: true
  }
}
layer {
  name: "layer_128_3_relu2"
  type: "ReLU"
  bottom: "layer_128_3_conv1"
  top: "layer_128_3_conv1"
}
layer {
  name: "layer_128_3_conv2"
  type: "Convolution"
  bottom: "layer_128_3_conv1"
  top: "layer_128_3_conv2"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  convolution_param {
    num_output: 128
    bias_term: false
    pad: 1
    kernel_size: 3
    stride: 1
    weight_filler {
      type: "msra"
    }
    bias_filler {
      type: "constant"
      value: 0.0
    }
  }
}
layer {
  name: "layer_128_3_bn3"
  type: "BatchNorm"
  bottom: "layer_128_3_conv2"
  top: "layer_128_3_conv2"
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
}
layer {
  name: "layer_128_3_scale3"
  type: "Scale"
  bottom: "layer_128_3_conv2"
  top: "layer_128_3_conv2"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  param {
    lr_mult: 2.0
    decay_mult: 1.0
  }
  scale_param {
    bias_term: true
  }
}
layer {
  name: "layer_128_3_relu3"
  type: "ReLU"
  bottom: "layer_128_3_conv2"
  top: "layer_128_3_conv2"
}
layer {
  name: "layer_128_3_conv3"
  type: "Convolution"
  bottom: "layer_128_3_conv2"
  top: "layer_128_3_conv3"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  convolution_param {
    num_output: 512
    bias_term: false
    pad: 0
    kernel_size: 1
    stride: 1
    weight_filler {
      type: "msra"
    }
    bias_filler {
      type: "constant"
      value: 0.0
    }
  }
}
layer {
  name: "layer_128_3_sum"
  type: "Eltwise"
  bottom: "layer_128_3_conv3"
  bottom: "layer_128_2_sum"
  top: "layer_128_3_sum"
}
layer {
  name: "layer_128_4_bn1"
  type: "BatchNorm"
  bottom: "layer_128_3_sum"
  top: "layer_128_4_bn1"
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
}
layer {
  name: "layer_128_4_scale1"
  type: "Scale"
  bottom: "layer_128_4_bn1"
  top: "layer_128_4_bn1"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  param {
    lr_mult: 2.0
    decay_mult: 1.0
  }
  scale_param {
    bias_term: true
  }
}
layer {
  name: "layer_128_4_relu1"
  type: "ReLU"
  bottom: "layer_128_4_bn1"
  top: "layer_128_4_bn1"
}
layer {
  name: "layer_128_4_conv1"
  type: "Convolution"
  bottom: "layer_128_4_bn1"
  top: "layer_128_4_conv1"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  convolution_param {
    num_output: 128
    bias_term: false
    pad: 0
    kernel_size: 1
    stride: 1
    weight_filler {
      type: "msra"
    }
    bias_filler {
      type: "constant"
      value: 0.0
    }
  }
}
layer {
  name: "layer_128_4_bn2"
  type: "BatchNorm"
  bottom: "layer_128_4_conv1"
  top: "layer_128_4_conv1"
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
}
layer {
  name: "layer_128_4_scale2"
  type: "Scale"
  bottom: "layer_128_4_conv1"
  top: "layer_128_4_conv1"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  param {
    lr_mult: 2.0
    decay_mult: 1.0
  }
  scale_param {
    bias_term: true
  }
}
layer {
  name: "layer_128_4_relu2"
  type: "ReLU"
  bottom: "layer_128_4_conv1"
  top: "layer_128_4_conv1"
}
layer {
  name: "layer_128_4_conv2"
  type: "Convolution"
  bottom: "layer_128_4_conv1"
  top: "layer_128_4_conv2"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  convolution_param {
    num_output: 128
    bias_term: false
    pad: 1
    kernel_size: 3
    stride: 1
    weight_filler {
      type: "msra"
    }
    bias_filler {
      type: "constant"
      value: 0.0
    }
  }
}
layer {
  name: "layer_128_4_bn3"
  type: "BatchNorm"
  bottom: "layer_128_4_conv2"
  top: "layer_128_4_conv2"
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
}
layer {
  name: "layer_128_4_scale3"
  type: "Scale"
  bottom: "layer_128_4_conv2"
  top: "layer_128_4_conv2"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  param {
    lr_mult: 2.0
    decay_mult: 1.0
  }
  scale_param {
    bias_term: true
  }
}
layer {
  name: "layer_128_4_relu3"
  type: "ReLU"
  bottom: "layer_128_4_conv2"
  top: "layer_128_4_conv2"
}
layer {
  name: "layer_128_4_conv3"
  type: "Convolution"
  bottom: "layer_128_4_conv2"
  top: "layer_128_4_conv3"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  convolution_param {
    num_output: 512
    bias_term: false
    pad: 0
    kernel_size: 1
    stride: 1
    weight_filler {
      type: "msra"
    }
    bias_filler {
      type: "constant"
      value: 0.0
    }
  }
}
layer {
  name: "layer_128_4_sum"
  type: "Eltwise"
  bottom: "layer_128_4_conv3"
  bottom: "layer_128_3_sum"
  top: "layer_128_4_sum"
}

layer {
  name: "last_bn"
  type: "BatchNorm"
  bottom: "layer_128_4_sum"
  top: "layer_128_4_sum"
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
}
layer {
  name: "last_scale"
  type: "Scale"
  bottom: "layer_128_4_sum"
  top: "layer_128_4_sum"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  param {
    lr_mult: 2.0
    decay_mult: 1.0
  }
  scale_param {
    bias_term: true
  }
}
layer {
  name: "last_relu"
  type: "ReLU"
  bottom: "layer_128_4_sum"
  top: "layer_128_4_sum"
}
layer {
  name: "dropout"
  type: "Dropout"
  bottom: "layer_128_4_sum"
  top: "layer_128_4_sum"
  dropout_param{
    dropout_ratio: 0.7
  }
}
layer {
    name: "permuted_data"
    type: "Permute"
    bottom: "layer_128_4_sum"
    top: "permuted_data"
    permute_param {
        order: 3
        order: 0
        order: 1
        order: 2
    }
}

layer {
  name: "lstm-reverse1"
  type: "Reverse"
  bottom: "permuted_data"
  top: "rlstm_input"
  reverse_param {
    axis: 0
  }
}
layer {
    name: "lstm2x"
    type: "LSTM"
    bottom: "rlstm_input"
    bottom: "indicator"
    top: "lstm2x"
    recurrent_param {
        num_output: 100
        weight_filler {
          type: "xavier"
        }
        bias_filler {
          type: "constant"
          value: 0
        }
    }
}
layer {
  name: "lstm-reverse2"
  type: "Reverse"
  bottom: "lstm2x"
  top: "rlstmx"
  reverse_param {
    axis: 0
  }
}

layer {
    name: "lstm1x"
    type: "LSTM"
    bottom: "permuted_data"
    bottom: "indicator"
    top: "lstm1x"
    recurrent_param {
        num_output: 100
        weight_filler {
          type: "xavier"
        }
        bias_filler {
          type: "constant"
          value: 0
        }
    }
}
layer {
  name: "merge_lstm_rlstmx"
  type: "Concat"
  bottom: "lstm1x"
  bottom: "rlstmx"
  top: "merge_lstm_rlstmx"
  concat_param {
    axis: 2
  }
}
layer {
  name: "fc1x"
  type: "InnerProduct"
  bottom: "merge_lstm_rlstmx"
  top: "fc1x"
  param {
    lr_mult: 1
    decay_mult: 1
  }
  param {
    lr_mult: 2
    decay_mult: 0
  }
  inner_product_param {
    axis: 2
    num_output: 33
    weight_filler {
      type: "xavier"
    }
    bias_filler {
      type: "constant"
      value: 0
    }
  }
}


layer {
  name: "ctc_loss"
  type: "CtcLoss"
  bottom: "fc1x"
  bottom: "label"
  top: "ctc_loss"
  loss_weight: 1.0
  ctc_loss_param {
    blank_label: 32
    alphabet_size: 33
    time_step: 19
  }
}

layer {
  name: "permute_fc"
  type: "Permute"
  bottom: "fc1x"
  top: "premuted_fc"
  include {
    phase: TEST
  }
  permute_param {
    order: 1
    order: 0
    order: 2
  }
}


layer {
  name: "accuracy"
  type: "LabelsequenceAccuracy"
  bottom: "premuted_fc"
  bottom: "label"
  top: "accuracy"
  include {
    phase: TEST
  }
  labelsequence_accuracy_param {
    blank_label: 32
  }
}