train.prototxt

name: "resnet10_bilstm_ctc_net"
layer {
    name: "data"
    type: "HDF5Data"
    top: "data"
    top: "label"
    include {
        phase: TRAIN
    }
    hdf5_data_param {
        source: ""
        batch_size: 64
    }
}

layer {
    name: "data"
    type: "HDF5Data"
    top: "data"
    top: "label"
    include {
        phase: TEST
    }
    hdf5_data_param {
        source: ""
        batch_size: 32
    }
}

layer {
  name: "data_bn"
  type: "BatchNorm"
  bottom: "data"
  top: "data_bn"
}
layer {
  name: "data_scale"
  type: "Scale"
  bottom: "data_bn"
  top: "data_bn"
  param {
    lr_mult: 1
    decay_mult: 1
  }
  param {
    lr_mult: 2
    decay_mult: 1
  }
  scale_param {
    bias_term: true
  }
}
layer {
  name: "conv1"
  type: "Convolution"
  bottom: "data_bn"
  top: "conv1"
  param {
    lr_mult: 1
    decay_mult: 1
  }
  param {
    lr_mult: 2
    decay_mult: 1
  }
  convolution_param {
    num_output: 64
    pad: 1
    kernel_size: 5
    stride: 1
    weight_filler {
      type: "msra"
      variance_norm: FAN_OUT
    }
    bias_filler {
      type: "constant"
      value: 0
    }
  }
}
layer {
  name: "conv1_bn"
  type: "BatchNorm"
  bottom: "conv1"
  top: "conv1"
}
layer {
  name: "conv1_scale"
  type: "Scale"
  bottom: "conv1"
  top: "conv1"
  param {
    lr_mult: 1
    decay_mult: 1
  }
  param {
    lr_mult: 2
    decay_mult: 1
  }
  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: 1
  }
}
layer {
  name: "layer_64_1_conv1"
  type: "Convolution"
  bottom: "conv1_pool"
  top: "layer_64_1_conv1"
  param {
    lr_mult: 1
    decay_mult: 1
  }
  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
    }
  }
}
layer {
  name: "layer_64_1_bn2"
  type: "BatchNorm"
  bottom: "layer_64_1_conv1"
  top: "layer_64_1_conv1"
}
layer {
  name: "layer_64_1_scale2"
  type: "Scale"
  bottom: "layer_64_1_conv1"
  top: "layer_64_1_conv1"
  param {
    lr_mult: 1
    decay_mult: 1
  }
  param {
    lr_mult: 2
    decay_mult: 1
  }
  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
    decay_mult: 1
  }
  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
    }
  }
}
layer {
  name: "layer_64_1_sum"
  type: "Eltwise"
  bottom: "layer_64_1_conv2"
  bottom: "conv1_pool"
  top: "layer_64_1_sum"
}
layer {
  name: "layer_128_1_bn1"
  type: "BatchNorm"
  bottom: "layer_64_1_sum"
  top: "layer_128_1_bn1"
}
layer {
  name: "layer_128_1_scale1"
  type: "Scale"
  bottom: "layer_128_1_bn1"
  top: "layer_128_1_bn1"
  param {
    lr_mult: 1
    decay_mult: 1
  }
  param {
    lr_mult: 2
    decay_mult: 1
  }
  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
    decay_mult: 1
  }
  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
    }
  }
}
layer {
  name: "layer_128_1_bn2"
  type: "BatchNorm"
  bottom: "layer_128_1_conv1"
  top: "layer_128_1_conv1"
}
layer {
  name: "layer_128_1_scale2"
  type: "Scale"
  bottom: "layer_128_1_conv1"
  top: "layer_128_1_conv1"
  param {
    lr_mult: 1
    decay_mult: 1
  }
  param {
    lr_mult: 2
    decay_mult: 1
  }
  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
    decay_mult: 1
  }
  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
    }
  }
}
layer {
  name: "layer_128_1_conv_expand"
  type: "Convolution"
  bottom: "layer_128_1_bn1"
  top: "layer_128_1_conv_expand"
  param {
    lr_mult: 1
    decay_mult: 1
  }
  convolution_param {
    num_output: 128
    bias_term: false
    pad: 0
    kernel_size: 1
    stride: 2
    weight_filler {
      type: "msra"
    }
    bias_filler {
      type: "constant"
      value: 0
    }
  }
}
layer {
  name: "layer_128_1_sum"
  type: "Eltwise"
  bottom: "layer_128_1_conv2"
  bottom: "layer_128_1_conv_expand"
  top: "layer_128_1_sum"
}
layer {
  name: "layer_256_1_bn1"
  type: "BatchNorm"
  bottom: "layer_128_1_sum"
  top: "layer_256_1_bn1"
}
layer {
  name: "layer_256_1_scale1"
  type: "Scale"
  bottom: "layer_256_1_bn1"
  top: "layer_256_1_bn1"
  param {
    lr_mult: 1
    decay_mult: 1
  }
  param {
    lr_mult: 2
    decay_mult: 1
  }
  scale_param {
    bias_term: true
  }
}
layer {
  name: "layer_256_1_relu1"
  type: "ReLU"
  bottom: "layer_256_1_bn1"
  top: "layer_256_1_bn1"
}
layer {
  name: "layer_256_1_conv1"
  type: "Convolution"
  bottom: "layer_256_1_bn1"
  top: "layer_256_1_conv1"
  param {
    lr_mult: 1
    decay_mult: 1
  }
  convolution_param {
    num_output: 256
    bias_term: false
    pad: 1
    kernel_size: 3
    stride: 2
    weight_filler {
      type: "msra"
    }
    bias_filler {
      type: "constant"
      value: 0
    }
  }
}
layer {
  name: "layer_256_1_bn2"
  type: "BatchNorm"
  bottom: "layer_256_1_conv1"
  top: "layer_256_1_conv1"
}
layer {
  name: "layer_256_1_scale2"
  type: "Scale"
  bottom: "layer_256_1_conv1"
  top: "layer_256_1_conv1"
  param {
    lr_mult: 1
    decay_mult: 1
  }
  param {
    lr_mult: 2
    decay_mult: 1
  }
  scale_param {
    bias_term: true
  }
}
layer {
  name: "layer_256_1_relu2"
  type: "ReLU"
  bottom: "layer_256_1_conv1"
  top: "layer_256_1_conv1"
}
layer {
  name: "layer_256_1_conv2"
  type: "Convolution"
  bottom: "layer_256_1_conv1"
  top: "layer_256_1_conv2"
  param {
    lr_mult: 1
    decay_mult: 1
  }
  convolution_param {
    num_output: 256
    bias_term: false
    pad: 1
    kernel_size: 3
    stride: 1
    weight_filler {
      type: "msra"
    }
    bias_filler {
      type: "constant"
      value: 0
    }
  }
}
layer {
  name: "layer_256_1_conv_expand"
  type: "Convolution"
  bottom: "layer_256_1_bn1"
  top: "layer_256_1_conv_expand"
  param {
    lr_mult: 1
    decay_mult: 1
  }
  convolution_param {
    num_output: 256
    bias_term: false
    pad: 0
    kernel_size: 1
    stride: 2
    weight_filler {
      type: "msra"
    }
    bias_filler {
      type: "constant"
      value: 0
    }
  }
}
layer {
  name: "layer_256_1_sum"
  type: "Eltwise"
  bottom: "layer_256_1_conv2"
  bottom: "layer_256_1_conv_expand"
  top: "layer_256_1_sum"
}
layer {
  name: "layer_512_1_bn1"
  type: "BatchNorm"
  bottom: "layer_256_1_sum"
  top: "layer_512_1_bn1"
}
layer {
  name: "layer_512_1_scale1"
  type: "Scale"
  bottom: "layer_512_1_bn1"
  top: "layer_512_1_bn1"
  param {
    lr_mult: 1
    decay_mult: 1
  }
  param {
    lr_mult: 2
    decay_mult: 1
  }
  scale_param {
    bias_term: true
  }
}
layer {
  name: "layer_512_1_relu1"
  type: "ReLU"
  bottom: "layer_512_1_bn1"
  top: "layer_512_1_bn1"
}
layer {
  name: "layer_512_1_conv1"
  type: "Convolution"
  bottom: "layer_512_1_bn1"
  top: "layer_512_1_conv1"
  param {
    lr_mult: 1
    decay_mult: 1
  }
  convolution_param {
    num_output: 512
    bias_term: false
    pad: 1
    kernel_size: 3
    stride: 2
    weight_filler {
      type: "msra"
    }
    bias_filler {
      type: "constant"
      value: 0
    }
  }
}
layer {
  name: "layer_512_1_bn2"
  type: "BatchNorm"
  bottom: "layer_512_1_conv1"
  top: "layer_512_1_conv1"
}
layer {
  name: "layer_512_1_scale2"
  type: "Scale"
  bottom: "layer_512_1_conv1"
  top: "layer_512_1_conv1"
  param {
    lr_mult: 1
    decay_mult: 1
  }
  param {
    lr_mult: 2
    decay_mult: 1
  }
  scale_param {
    bias_term: true
  }
}
layer {
  name: "layer_512_1_relu2"
  type: "ReLU"
  bottom: "layer_512_1_conv1"
  top: "layer_512_1_conv1"
}
layer {
  name: "layer_512_1_conv2"
  type: "Convolution"
  bottom: "layer_512_1_conv1"
  top: "layer_512_1_conv2"
  param {
    lr_mult: 1
    decay_mult: 1
  }
  convolution_param {
    num_output: 512
    bias_term: false
    pad: 1
    kernel_size: 3
    stride: 1
    weight_filler {
      type: "msra"
    }
    bias_filler {
      type: "constant"
      value: 0
    }
  }
}
layer {
  name: "layer_512_1_conv_expand"
  type: "Convolution"
  bottom: "layer_512_1_bn1"
  top: "layer_512_1_conv_expand"
  param {
    lr_mult: 1
    decay_mult: 1
  }
  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
    }
  }
}
layer {
  name: "layer_512_1_sum"
  type: "Eltwise"
  bottom: "layer_512_1_conv2"
  bottom: "layer_512_1_conv_expand"
  top: "layer_512_1_sum"
}
layer {
  name: "last_bn"
  type: "BatchNorm"
  bottom: "layer_512_1_sum"
  top: "layer_512_1_sum"
}
layer {
  name: "last_scale"
  type: "Scale"
  bottom: "layer_512_1_sum"
  top: "layer_512_1_sum"
  param {
    lr_mult: 1
    decay_mult: 1
  }
  param {
    lr_mult: 2
    decay_mult: 1
  }
  scale_param {
    bias_term: true
  }
}
layer {
  name: "last_relu"
  type: "ReLU"
  bottom: "layer_512_1_sum"
  top: "layer_512_1_sum"
}











layer {
    name: "indicator"
    type: "ContinuationIndicator"
    top: "indicator"
    continuation_indicator_param {
        time_step:  15
        batch_size: 64
    }
	include {
        phase: TRAIN
    }
}

layer {
    name: "indicator"
    type: "ContinuationIndicator"
    top: "indicator"
    continuation_indicator_param {
        time_step:  15
        batch_size: 32
    }
	include {
        phase: TEST
    }
}


layer {
    name: "permuted_data"
    type: "Permute"
    bottom: "layer_512_1_sum"
    top: "permuted_data"
    permute_param {
        order: 3
        order: 0
        order: 1
        order: 2
    }
}

layer {
    name: "lstm1"
    type: "LSTM"
    bottom: "permuted_data"
    bottom: "indicator"
    top: "lstm1"
    recurrent_param {
        num_output: 100
        weight_filler {
          type: "xavier"
        }
        bias_filler {
          type: "constant"
          value: 0
        }
    }
}

layer {
    name: "lstm2"
    type: "LSTM"
    bottom: "lstm1"
    bottom: "indicator"
    top: "lstm2"
    recurrent_param {
        num_output: 100
        weight_filler {
          type: "xavier"
        }
        bias_filler {
          type: "constant"
          value: 0
        }
    }
}

layer {
  name: "fc1"
  type: "InnerProduct"
  bottom: "lstm2"
  top: "fc1"
  param {
    lr_mult: 1
    decay_mult: 1
  }
  param {
    lr_mult: 2
    decay_mult: 0
  }
  inner_product_param {
    num_output: 100
	  axis: 2
    weight_filler {
      type: "xavier"
    }
    bias_filler {
      type: "constant"
      value: 0
    }
  }
}


layer {
  name: "fc2"
  type: "InnerProduct"
  bottom: "fc1"
  top: "fc2"
  param {
    lr_mult: 1
    decay_mult: 1
  }
  param {
    lr_mult: 2
    decay_mult: 0
  }
  inner_product_param {
    num_output: 11
	  axis: 2
    weight_filler {
      type: "xavier"
    }
    bias_filler {
      type: "constant"
      value: 0
    }
  }
}

layer {
  name: "rev1"
  type: "Reverse"
  bottom: "permuted_data"
  top: "rev_data"
}


layer {
    name: "lstm3"
    type: "LSTM"
    bottom: "rev_data"
    bottom: "indicator"
    top: "lstm3"
    recurrent_param {
        num_output: 100
        weight_filler {
          type: "xavier"
        }
        bias_filler {
          type: "constant"
          value: 0
        }
    }
}

layer {
    name: "lstm4"
    type: "LSTM"
    bottom: "lstm3"
    bottom: "indicator"
    top: "lstm4"
    recurrent_param {
        num_output: 100
        weight_filler {
          type: "xavier"
        }
        bias_filler {
          type: "constant"
          value: 0
        }
    }
}

layer {
  name: "fc3"
  type: "InnerProduct"
  bottom: "lstm4"
  top: "fc3"
  param {
    lr_mult: 1
    decay_mult: 1
  }
  param {
    lr_mult: 2
    decay_mult: 0
  }
  inner_product_param {
    num_output: 100
	  axis: 2
    weight_filler {
      type: "xavier"
    }
    bias_filler {
      type: "constant"
      value: 0
    }
  }
}

layer {
  name: "fc4"
  type: "InnerProduct"
  bottom: "fc3"
  top: "fc4"
  param {
    lr_mult: 1
    decay_mult: 1
  }
  param {
    lr_mult: 2
    decay_mult: 0
  }
  inner_product_param {
    num_output: 11
	  axis: 2
    weight_filler {
      type: "xavier"
    }
    bias_filler {
      type: "constant"
      value: 0
    }
  }
}

layer {
  name: "eltwise-sum"
  type: "Eltwise"
  bottom: "fc2"
  bottom: "fc4"
  eltwise_param { operation: SUM }
  top: "sum"
}

layer {
	name: "ctc_loss"
	type: "CtcLoss"
	bottom: "sum"
	bottom: "label"
	top: "ctc_loss"
	loss_weight: 1.0
	ctc_loss_param {
		blank_label: 10
		alphabet_size: 11
    time_step: 15
	}
}

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


layer {
  name: "accuracy"
  type: "SequenceAccuracy"
  bottom: "premuted_fc"
  bottom: "label"
  top: "accuracy"
  include {
    phase: TEST
  }
  Sequence_accuracy_param {
    blank_label: 10
	#accuracy_type: 1
  }
}