train.prototxt

name: "MSCNN"

layer {
  name: "data"
  type: "CrowdData"
  top: "data"
  top: "label"
  crowd_data_param {
    source: "train_aug.txt"
    batch_size: 1
    mirror: True
    shuffle: True
    is_color: False
  }
}
layer {
  name: "conv1"
  type: "Convolution"
  bottom: "data"
  top: "conv1"
  param {
    lr_mult: 1
  }
  param {
    lr_mult: 2
  }
  convolution_param {
    num_output: 64
    pad: 4
    kernel_size: 9
    stride: 1
    weight_filler {
      type: "gaussian"
      std: 0.01000000
    }
    bias_filler {
      type: "constant"
      value: 0
    }
  }
}
layer {
  name: "relu1"
  type: "ReLU"
  bottom: "conv1"
  top: "conv1"
}
layer {
  name: "conv2_1"
  type: "Convolution"
  bottom: "conv1"
  top: "conv2_1"
  param {
    lr_mult: 1
  }
  param {
    lr_mult: 2
  }
  convolution_param {
    num_output: 16
    pad: 4
    kernel_size: 9
    stride: 1
    weight_filler {
      type: "gaussian"
      std: 0.01000000
    }
    bias_filler {
      type: "constant"
      value: 0
    }
  }
}
layer {
  name: "conv2_2"
  type: "Convolution"
  bottom: "conv1"
  top: "conv2_2"
  param {
    lr_mult: 1
  }
  param {
    lr_mult: 2
  }
  convolution_param {
    num_output: 16
    pad: 3
    kernel_size: 7
    stride: 1
    weight_filler {
      type: "gaussian"
      std: 0.01000000
    }
    bias_filler {
      type: "constant"
      value: 0
    }
  }
}
layer {
  name: "conv2_3"
  type: "Convolution"
  bottom: "conv1"
  top: "conv2_3"
  param {
    lr_mult: 1
  }
  param {
    lr_mult: 2
  }
  convolution_param {
    num_output: 16
    pad: 2
    kernel_size: 5
    stride: 1
    weight_filler {
      type: "gaussian"
      std: 0.01000000
    }
    bias_filler {
      type: "constant"
      value: 0
    }
  }
}
layer {
  name: "conv2_4"
  type: "Convolution"
  bottom: "conv1"
  top: "conv2_4"
  param {
    lr_mult: 1
  }
  param {
    lr_mult: 2
  }
  convolution_param {
    num_output: 16
    pad: 1
    kernel_size: 3
    stride: 1
    weight_filler {
      type: "gaussian"
      std: 0.01000000
    }
    bias_filler {
      type: "constant"
      value: 0
    }
  }
}
layer {
  name: "conv2"
  type: "Concat"
  bottom: "conv2_1"
  bottom: "conv2_2"
  bottom: "conv2_3"
  bottom: "conv2_4"
  top: "conv2"
  concat_param {
  axis: 1
  }
}
layer {
  name: "relu2"
  type: "ReLU"
  bottom: "conv2"
  top: "conv2"
}
layer {
  name: "pool1"
  type: "Pooling"
  bottom: "conv2"
  top: "pool1"
  pooling_param {
    pool: MAX
    kernel_size: 2
    stride: 2
  }
}
layer {
  name: "conv3_1"
  type: "Convolution"
  bottom: "pool1"
  top: "conv3_1"
  param {
    lr_mult: 1
  }
  param {
    lr_mult: 2
  }
  convolution_param {
    num_output: 32
    pad: 4
    kernel_size: 9
    stride: 1
    weight_filler {
      type: "gaussian"
      std: 0.01000000
    }
    bias_filler {
      type: "constant"
      value: 0
    }
  }
}
layer {
  name: "conv3_2"
  type: "Convolution"
  bottom: "pool1"
  top: "conv3_2"
  param {
    lr_mult: 1
  }
  param {
    lr_mult: 2
  }
  convolution_param {
    num_output: 32
    pad: 3
    kernel_size: 7
    stride: 1
    weight_filler {
      type: "gaussian"
      std: 0.01000000
    }
    bias_filler {
      type: "constant"
      value: 0
    }
  }
}
layer {
  name: "conv3_3"
  type: "Convolution"
  bottom: "pool1"
  top: "conv3_3"
  param {
    lr_mult: 1
  }
  param {
    lr_mult: 2
  }
  convolution_param {
    num_output: 32
    pad: 2
    kernel_size: 5
    stride: 1
    weight_filler {
      type: "gaussian"
      std: 0.01000000
    }
    bias_filler {
      type: "constant"
      value: 0
    }
  }
}
layer {
  name: "conv3_4"
  type: "Convolution"
  bottom: "pool1"
  top: "conv3_4"
  param {
    lr_mult: 1
  }
  param {
    lr_mult: 2
  }
  convolution_param {
    num_output: 32
    pad: 1
    kernel_size: 3
    stride: 1
    weight_filler {
      type: "gaussian"
      std: 0.01000000
    }
    bias_filler {
      type: "constant"
      value: 0
    }
  }
}
layer {
  name: "conv3"
  type: "Concat"
  bottom: "conv3_1"
  bottom: "conv3_2"
  bottom: "conv3_3"
  bottom: "conv3_4"
  top: "conv3"
  concat_param {
  axis: 1
  }
}
layer {
  name: "relu3"
  type: "ReLU"
  bottom: "conv3"
  top: "conv3"
}
layer {
  name: "conv4_1"
  type: "Convolution"
  bottom: "conv3"
  top: "conv4_1"
  param {
    lr_mult: 1
  }
  param {
    lr_mult: 2
  }
  convolution_param {
    num_output: 32
    pad: 4
    kernel_size: 9
    stride: 1
    weight_filler {
      type: "gaussian"
      std: 0.01000000
    }
    bias_filler {
      type: "constant"
      value: 0
    }
  }
}
layer {
  name: "conv4_2"
  type: "Convolution"
  bottom: "conv3"
  top: "conv4_2"
  param {
    lr_mult: 1
  }
  param {
    lr_mult: 2
  }
  convolution_param {
    num_output: 32
    pad: 3
    kernel_size: 7
    stride: 1
    weight_filler {
      type: "gaussian"
      std: 0.01000000
    }
    bias_filler {
      type: "constant"
      value: 0
    }
  }
}
layer {
  name: "conv4_3"
  type: "Convolution"
  bottom: "conv3"
  top: "conv4_3"
  param {
    lr_mult: 1
  }
  param {
    lr_mult: 2
  }
  convolution_param {
    num_output: 32
    pad: 2
    kernel_size: 5
    stride: 1
    weight_filler {
      type: "gaussian"
      std: 0.01000000
    }
    bias_filler {
      type: "constant"
      value: 0
    }
  }
}
layer {
  name: "conv4_4"
  type: "Convolution"
  bottom: "conv3"
  top: "conv4_4"
  param {
    lr_mult: 1
  }
  param {
    lr_mult: 2
  }
  convolution_param {
    num_output: 32
    pad: 1
    kernel_size: 3
    stride: 1
    weight_filler {
      type: "gaussian"
      std: 0.01000000
    }
    bias_filler {
      type: "constant"
      value: 0
    }
  }
}
layer {
  name: "conv4"
  type: "Concat"
  bottom: "conv4_1"
  bottom: "conv4_2"
  bottom: "conv4_3"
  bottom: "conv4_4"
  top: "conv4"
  concat_param {
   axis: 1
  }
}
layer {
  name: "relu4"
  type: "ReLU"
  bottom: "conv4"
  top: "conv4"
}
layer {
  name: "pool2"
  type: "Pooling"
  bottom: "conv4"
  top: "pool2"
  pooling_param {
    pool: MAX
    kernel_size: 2
    stride: 2
  }
}
layer {
  name: "conv5_1"
  type: "Convolution"
  bottom: "pool2"
  top: "conv5_1"
  param {
    lr_mult: 1
  }
  param {
    lr_mult: 2
  }
  convolution_param {
    num_output: 64
    pad: 3
    kernel_size: 7
    stride: 1
    weight_filler {
      type: "gaussian"
      std: 0.01000000
    }
    bias_filler {
      type: "constant"
      value: 0
    }
  }
}
layer {
  name: "conv5_2"
  type: "Convolution"
  bottom: "pool2"
  top: "conv5_2"
  param {
    lr_mult: 1
  }
  param {
    lr_mult: 2
  }
  convolution_param {
    num_output: 64
    pad: 2
    kernel_size: 5
    stride: 1
    weight_filler {
      type: "gaussian"
      std: 0.01000000
    }
    bias_filler {
      type: "constant"
      value: 0
    }
  }
}
layer {
  name: "conv5_3"
  type: "Convolution"
  bottom: "pool2"
  top: "conv5_3"
  param {
    lr_mult: 1
  }
  param {
    lr_mult: 2
  }
  convolution_param {
    num_output: 64
    pad: 1
    kernel_size: 3
    stride: 1
    weight_filler {
      type: "gaussian"
      std: 0.01000000
    }
    bias_filler {
      type: "constant"
      value: 0
    }
  }
}
layer {
  name: "conv5"
  type: "Concat"
  bottom: "conv5_1"
  bottom: "conv5_2"
  bottom: "conv5_3"
  top: "conv5"
  concat_param {
   axis: 1
  }
}
layer {
  name: "relu5"
  type: "ReLU"
  bottom: "conv5"
  top: "conv5"
}
layer {
  name: "conv6_1"
  type: "Convolution"
  bottom: "conv5"
  top: "conv6_1"
  param {
    lr_mult: 1
  }
  param {
    lr_mult: 2
  }
  convolution_param {
    num_output: 64
    pad: 3
    kernel_size: 7
    stride: 1
    weight_filler {
      type: "gaussian"
      std: 0.01000000
    }
    bias_filler {
      type: "constant"
      value: 0
    }
  }
}
layer {
  name: "conv6_2"
  type: "Convolution"
  bottom: "conv5"
  top: "conv6_2"
  param {
    lr_mult: 1
  }
  param {
    lr_mult: 2
  }
  convolution_param {
    num_output: 64
    pad: 2
    kernel_size: 5
    stride: 1
    weight_filler {
      type: "gaussian"
      std: 0.01000000
    }
    bias_filler {
      type: "constant"
      value: 0
    }
  }
}
layer {
  name: "conv6_3"
  type: "Convolution"
  bottom: "conv5"
  top: "conv6_3"
  param {
    lr_mult: 1
  }
  param {
    lr_mult: 2
  }
  convolution_param {
    num_output: 64
    pad: 1
    kernel_size: 3
    stride: 1
    weight_filler {
      type: "gaussian"
      std: 0.01000000
    }
    bias_filler {
      type: "constant"
      value: 0
    }
  }
}
layer {
  name: "conv6"
  type: "Concat"
  bottom: "conv6_1"
  bottom: "conv6_2"
  bottom: "conv6_3"
  top: "conv6"
  concat_param {
   axis: 1
  }
}
layer {
  name: "relu6"
  type: "ReLU"
  bottom: "conv6"
  top: "conv6"
}
layer {
  name: "conv"
  type: "Convolution"
  bottom: "conv6"
  top: "estdmap"
  convolution_param {
    num_output: 1
    kernel_size:1
    weight_filler {
      type: "gaussian"
      std: 0.01000000
    }
  }
}
layer {
  name: "loss"
  type: "EuclideanLoss"
  bottom: "estdmap"
  bottom: "label"
  top: "loss"
}