如何将 caffe prototxt 转换为 pytorch 模型？答案

【问题标题】：How to convert caffe prototxt to pytorch model?如何将 caffe prototxt 转换为 pytorch 模型？
【发布时间】：2018-05-21 07:30:50
【问题描述】：

到目前为止，我使用的是 pytorch-caffe-darknet-convert 存储库。在克服了许多问题（concat 和 eltwise 层不可转换）之后，我最终得到了一个看起来像暗网配置文件的东西：

python caffe2darknet.py my_prototxt.txt my_caffemodel.caffemodel new_net_file.cfg new_model.weights

有人知道如何将输出 new_net_file.cfg 转换为 pytorch 吗？或者还有另一种将 caffe prototxt 文件转换为 pytorch 的方法吗？
我希望有与caffe-tensorflow 相同的行为我将在下面发布我的 caffe prototxt 和输出 new_net_file.cfg 作为参考。

my_prototxt：

input: "data"
input_shape {
  dim: 1
  dim: 240
  dim: 144
  dim: 240
}

layer {
  name: "conv1_1"
  type: "Convolution"
  bottom: "data"
  top: "conv1_1"
  convolution_param {
    num_output: 16
    pad: 3
    pad: 3
    pad: 3
    kernel_size: 7
    kernel_size: 7
    kernel_size: 7
    stride: 2
    weight_filler {
      type: "xavier"
    }
    bias_filler {
      type: "constant"
    }
    engine: CUDNN
    axis: 1
  }
}
layer {
  name: "relu1_1"
  type: "ReLU"
  bottom: "conv1_1"
  top: "conv1_1"
}
layer {
  name: "reduction2_1"
  type: "Convolution"
  bottom: "conv1_1"
  top: "reduction2_1"
  convolution_param {
    num_output: 32
    bias_term: false
    pad: 0
    kernel_size: 1
    stride: 1
    weight_filler {
      type: "xavier"
    }
    bias_filler {
      type: "constant"
    }
  }
}
layer {
  name: "conv2_1"
  type: "Convolution"
  bottom: "conv1_1"
  top: "conv2_1"
  convolution_param {
    num_output: 32
    pad: 1
    pad: 1
    pad: 1
    kernel_size: 3
    kernel_size: 3
    kernel_size: 3
    stride: 1
    weight_filler {
      type: "xavier"
    }
    bias_filler {
      type: "constant"
    }
    engine: CUDNN
    axis: 1
  }
}
layer {
  name: "relu2_1"
  type: "ReLU"
  bottom: "conv2_1"
  top: "conv2_1"
}
layer {
  name: "conv2_2"
  type: "Convolution"
  bottom: "conv2_1"
  top: "conv2_2"
  convolution_param {
    num_output: 32
    pad: 1
    pad: 1
    pad: 1
    kernel_size: 3
    kernel_size: 3
    kernel_size: 3
    stride: 1
    weight_filler {
      type: "xavier"
    }
    bias_filler {
      type: "constant"
    }
    axis: 1
  }
}
layer {
  name: "res2_2"
  type: "Eltwise"
  bottom: "reduction2_1"
  bottom: "conv2_2"
  top: "res2_2"
  eltwise_param { operation: SUM }
}
layer {
  name: "add2_2"
  type: "ReLU"
  bottom: "res2_2"
  top: "res2_2"
}
layer {
  name: "pool2"
  type: "Pooling"
  bottom: "res2_2"
  top: "pool2"
  pooling_param {
    pool: MAX
    kernel_size: 2
    stride: 2
    engine: CUDNN
  }
}
[...] # I cropped it here, since file is too lengthy

（暗网）配置文件：

[net]
batch=1
channels=240
height=144
width=240

[convolutional]
filters=16
size=['7', '7', '7']
stride=2
pad=1
activation=relu

[convolutional]
filters=32
size=1
stride=1
pad=1
activation=linear

[route]
layers=-2

[convolutional]
filters=32
size=['3', '3', '3']
stride=1
pad=1
activation=relu

[convolutional]
filters=32
size=['3', '3', '3']
stride=1
pad=1
activation=linear

[shortcut]
from=-4
activation=relu

[maxpool]
size=2
stride=2

[...] # I cropped it here, since file is too lengthy

【问题讨论】：

我没有仔细看，但你可以转换Caffe to Caffe2，Caffe2 to ONNX, then to PyTorch。或者，从Caffe2 to ONNX 查看此引用，然后查看Pytorch。不过，一定有更好的方法。
或者，您可以从其他人的暗网 pytorch 实施开始，例如：github.com/ayooshkathuria/pytorch-yolo-v3/blob/master/…

标签： python pytorch caffe torch darknet

【解决方案1】：

您可以使用以下库之一：

用法

转化

python caffe2pth_convertor.py \
--prototxt=YOUT_PROTOTXT_PATH \
--caffemodel=YOUT_CAFFEMODEL_PATH \
--pthmodel=OUTPUT_PTHMODEL_PATH

在 Pytorch 中使用模型

from caffe2pth.caffenet import *

net = CaffeNet(YOUT_PROTOTXT_PATH)
net.load_state_dict(torch.load(OUTPUT_PTHMODEL_PATH))

【讨论】：