name: "VPGNet-noVP"

input: "data"
input_shape {
  dim: 1
  dim: 3
  dim: 480
  dim: 640
}

layer {
  name: "L0"
  type: "Convolution"
  bottom: "data"
  top: "L0"
  param {
    lr_mult: 1
    decay_mult: 1
  }
  param {
    lr_mult: 2
    decay_mult: 0
  }
  convolution_param {
    num_output: 96
    kernel_size: 11
    stride: 4
    weight_filler {
      type: "gaussian"
      std: 0.01
    }
    bias_filler {
      type: "constant"
      value: 1
    }
  }
}
layer {
  name: "relu1"
  type: "ReLU"
  bottom: "L0"
  top: "L0"
}
layer {
  name: "norm1"
  type: "LRN"
  bottom: "L0"
  top: "norm1"
  lrn_param {
    k: 2
    local_size: 5
    alpha: 0.0005
    beta: 0.75
  }
}
layer {
  name: "pool1"
  type: "Pooling"
  bottom: "norm1"
  top: "pool1"
  pooling_param {
    pool: MAX
    kernel_size: 3
    stride: 2
  }
}
layer {
  name: "L1"
  type: "Convolution"
  bottom: "pool1"
  top: "L1"
  param {
    lr_mult: 1
    decay_mult: 1
  }
  param {
    lr_mult: 2
    decay_mult: 0
  }
  convolution_param {
    num_output: 256
    pad: 2
    kernel_size: 5
    group: 1
    weight_filler {
      type: "gaussian"
      std: 0.01
    }
    bias_filler {
      type: "constant"
      value: 1
    }
  }
}
layer {
  name: "relu2"
  type: "ReLU"
  bottom: "L1"
  top: "L1"
}
layer {
  name: "norm2"
  type: "LRN"
  bottom: "L1"
  top: "norm2"
  lrn_param {
    k: 8
    local_size: 5
    alpha: 0.0005
    beta: 0.75
  }
}
layer {
  name: "pool2"
  type: "Pooling"
  bottom: "norm2"
  top: "pool2"
  pooling_param {
    pool: MAX
    kernel_size: 3
    stride: 2
  }
}
layer {
  name: "L2"
  type: "Convolution"
  bottom: "pool2"
  top: "L2"
  param {
    lr_mult: 1
    decay_mult: 1
  }
  param {
    lr_mult: 2
    decay_mult: 0
  }
  convolution_param {
    num_output: 384
    pad: 1
    kernel_size: 3
    weight_filler {
      type: "gaussian"
      std: 0.01
    }
    bias_filler {
      type: "constant"
      value: 0.1
    }
  }
}
layer {
  name: "relu3"
  type: "ReLU"
  bottom: "L2"
  top: "L2"
}
layer {
  name: "L3"
  type: "Convolution"
  bottom: "L2"
  top: "L3"
  param {
    lr_mult: 1
    decay_mult: 1
  }
  param {
    lr_mult: 2
    decay_mult: 0
  }
  convolution_param {
    num_output: 384
    pad: 1
    kernel_size: 3
    group: 1
    weight_filler {
      type: "gaussian"
      std: 0.01
    }
    bias_filler {
      type: "constant"
      value: 0.1
    }
  }
}
layer {
  name: "relu4"
  type: "ReLU"
  bottom: "L3"
  top: "L3"
}
layer {
  name: "L4"
  type: "Convolution"
  bottom: "L3"
  top: "L4"
  param {
    lr_mult: 1
    decay_mult: 1
  }
  param {
    lr_mult: 2
    decay_mult: 0
  }
  convolution_param {
    num_output: 384
    pad: 1
    kernel_size: 3
    group: 1
    weight_filler {
      type: "gaussian"
      std: 0.01
    }
    bias_filler {
      type: "constant"
      value: 0.1
    }
  }
}
layer {
  name: "relu5"
  type: "ReLU"
  bottom: "L4"
  top: "L4"
}
layer {
  name: "pool5"
  type: "Pooling"
  bottom: "L4"
  top: "pool5"
  pooling_param {
    pool: MAX
    kernel_size: 3
    stride: 2
  }
}

layer {
  name: "L5"
  type: "Convolution"
  bottom: "pool5"
  top: "L5"
  param {
    lr_mult: 1
    decay_mult: 1
  }
  param {
    lr_mult: 2
    decay_mult: 0
  }
  convolution_param {
    num_output: 4096
    kernel_size: 6
    pad: 3
    weight_filler {
      type: "gaussian"
      std: 0.01
    }
    bias_filler {
      type: "constant"
      value: 1
    }
  }
}
layer {
  name: "relu6"
  type: "ReLU"
  bottom: "L5"
  top: "L5"
}

layer {
  name: "L6a"
  type: "Convolution"
  bottom: "L5"
  top: "L6a"
  param {
    lr_mult: 5
    decay_mult: 0.1
  }
  param {
    lr_mult: 2
    decay_mult: 0
  }
  convolution_param {
    num_output: 4096
    kernel_size: 1
    weight_filler {
      type: "gaussian"
      std: 0.01
    }
    bias_filler {
      type: "constant"
      value: 1
    }
  }
}
layer {
  name: "relu7a"
  type: "ReLU"
  bottom: "L6a"
  top: "L6a"
}

layer {
  name: "L6b"
  type: "Convolution"
  bottom: "L5"
  top: "L6b"
  param {
    lr_mult: 5
    decay_mult: 0.1
  }
  param {
    lr_mult: 2
    decay_mult: 0
  }
  convolution_param {
    num_output: 4096
    kernel_size: 1
    weight_filler {
      type: "gaussian"
      std: 0.01
    }
    bias_filler {
      type: "constant"
      value: 1
    }
  }
}
layer {
  name: "relu7b"
  type: "ReLU"
  bottom: "L6b"
  top: "L6b"
}

layer {
  name: "L6c"
  type: "Convolution"
  bottom: "L5"
  top: "L6c"
  param {
    lr_mult: 5
    decay_mult: 0.1
  }
  param {
    lr_mult: 2
    decay_mult: 0
  }
  convolution_param {
    num_output: 4096
    kernel_size: 1
    weight_filler {
      type: "gaussian"
      std: 0.01
    }
    bias_filler {
      type: "constant"
      value: 1
    }
  }
}
layer {
  name: "relu7c"
  type: "ReLU"
  bottom: "L6c"
  top: "L6c"
}

layer {
  name: "bb-output"
  type: "Convolution"
  bottom: "L6a"
  top: "bb-output"
  param {
    lr_mult: 20
    decay_mult: 0.1
  }
  param {
    lr_mult: 20
    decay_mult: 0
  }
  convolution_param {
    num_output: 256
    kernel_size: 1
    weight_filler {
      type: "gaussian"
      std: 0.01
    }
    bias_filler {
      type: "constant"
      value: 0.0
    }
  }
}

layer {
  name: "pixel-conv"
  type: "Convolution"
  bottom: "L6b"
  top: "pixel-conv"
  param {
    lr_mult: 1
    decay_mult: 1
  }
  param {
    lr_mult: 2
    decay_mult: 0
  }
  convolution_param {
    num_output: 128
    kernel_size: 1
    weight_filler {
      type: "gaussian"
      std: 0.01
    }
    bias_filler {
      type: "constant"
      value: 0.0
    }
  }
}

layer {
  name: "type-conv"
  type: "Convolution"
  bottom: "L6c"
  top: "type-conv"
  param {
    lr_mult: 1
    decay_mult: 1
  }
  param {
    lr_mult: 2
    decay_mult: 0
  }
  convolution_param {
    num_output: 1024
    kernel_size: 1
    weight_filler {
      type: "gaussian"
      std: 0.01
    }
    bias_filler {
      type: "constant"
      value: 0.0
    }
  }
}

layer {
  name: "pixel-tile"
  type: "Tiling"
  bottom: "pixel-conv"
  top: "pixel-conv-tiled"
  tiling_param {
    tile_dim: 8
  }
}

layer {
  name: "bb-tile"
  type: "Tiling"
  bottom: "bb-output"
  top: "bb-output-tiled"
  tiling_param {
    tile_dim: 8
  }
}

layer {
  name: "type-tile"
  type: "Tiling"
  bottom: "type-conv"
  top: "type-conv-tiled"
  tiling_param {
    tile_dim: 4
  }
}

# Pixel level softmax prediction.
layer {
  name: "binary-mask"
  type: "Softmax"
  bottom: "pixel-conv-tiled"
  top: "binary-mask"
}

# Pixel level softmax prediction.
layer {
  name: "multi-label"
  type: "Softmax"
  bottom: "type-conv-tiled"
  top: "multi-label"
}