我正在尝试将Prototxt文件转换为cfg。在处理过程中，我在github https://github.com/marvis/pytorch-caffe-darknet-convert中找到了它，将prototxt转换为cfg很有用，但它仅用于单输入转换。就我而言，prototxt具有四个输入，因此我无法将其转换为cfg。是否有任何过程或方法可以将具有多个输入的prototxt转换为等效的cfg。以下是我的原型：
name: "iTracker_test"
input: "image_left"
input_dim: 256
input_dim: 3
input_dim: 224
input_dim: 224
input: "image_right"
input_dim: 256
input_dim: 3
input_dim: 224
input_dim: 224
input: "image_face"
input_dim: 256
input_dim: 3
input_dim: 224
input_dim: 224
input: "facegrid"
input_dim: 256
input_dim: 625
input_dim: 1
input_dim: 1
layer {
name: "conv1"
type: "Convolution"
bottom: "image_left"
top: "conv1"
param {
name: "conv1_w"
lr_mult: 1
decay_mult: 1
}
param {
name: "conv1_b"
lr_mult: 2
decay_mult: 0
}
convolution_param {
num_output: 96
kernel_size: 11
stride: 4
}
}
layer {
name: "relu1"
type: "ReLU"
bottom: "conv1"
top: "conv1"
}
layer {
name: "pool1"
type: "Pooling"
bottom: "conv1"
top: "pool1"
pooling_param {
pool: MAX
kernel_size: 3
stride: 2
}
}
layer {
name: "norm1"
type: "LRN"
bottom: "pool1"
top: "norm1"
lrn_param {
local_size: 5
alpha: 0.0001
beta: 0.75
}
}
layer {
name: "conv2"
type: "Convolution"
bottom: "norm1"
top: "conv2"
param {
name: "conv2_w"
lr_mult: 1
decay_mult: 1
}
param {
name: "conv2_b"
lr_mult: 2
decay_mult: 0
}
convolution_param {
num_output: 256
pad: 2
kernel_size: 5
group: 2
}
}
layer {
name: "relu2"
type: "ReLU"
bottom: "conv2"
top: "conv2"
}
layer {
name: "pool2"
type: "Pooling"
bottom: "conv2"
top: "pool2"
pooling_param {
pool: MAX
kernel_size: 3
stride: 2
}
}
layer {
name: "norm2"
type: "LRN"
bottom: "pool2"
top: "norm2"
lrn_param {
local_size: 5
alpha: 0.0001
beta: 0.75
}
}
layer {
name: "conv3"
type: "Convolution"
bottom: "norm2"
top: "conv3"
param {
name: "conv3_w"
lr_mult: 1
decay_mult: 1
}
param {
name: "conv3_b"
lr_mult: 2
decay_mult: 0
}
convolution_param {
num_output: 384
pad: 1
kernel_size: 3
}
}
layer {
name: "relu3"
type: "ReLU"
bottom: "conv3"
top: "conv3"
}
layer {
name: "conv4_l"
type: "Convolution"
bottom: "conv3"
top: "conv4_l"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 2
decay_mult: 0
}
convolution_param {
num_output: 64
pad: 0
kernel_size: 1
}
}
layer {
name: "relu4_l"
type: "ReLU"
bottom: "conv4_l"
top: "conv4_l"
}
layer {
name: "conv1_r"
type: "Convolution"
bottom: "image_right"
top: "conv1_r"
param {
name: "conv1_w"
lr_mult: 1
decay_mult: 1
}
param {
name: "conv1_b"
lr_mult: 2
decay_mult: 0
}
convolution_param {
num_output: 96
kernel_size: 11
stride: 4
}
}
layer {
name: "relu1_r"
type: "ReLU"
bottom: "conv1_r"
top: "conv1_r"
}
layer {
name: "pool1_r"
type: "Pooling"
bottom: "conv1_r"
top: "pool1_r"
pooling_param {
pool: MAX
kernel_size: 3
stride: 2
}
}
layer {
name: "norm1_r"
type: "LRN"
bottom: "pool1_r"
top: "norm1_r"
lrn_param {
local_size: 5
alpha: 0.0001
beta: 0.75
}
}
layer {
name: "conv2_r"
type: "Convolution"
bottom: "norm1_r"
top: "conv2_r"
param {
name: "conv2_w"
lr_mult: 1
decay_mult: 1
}
param {
name: "conv2_b"
lr_mult: 2
decay_mult: 0
}
convolution_param {
num_output: 256
pad: 2
kernel_size: 5
group: 2
}
}
layer {
name: "relu2_r"
type: "ReLU"
bottom: "conv2_r"
top: "conv2_r"
}
layer {
name: "pool2_r"
type: "Pooling"
bottom: "conv2_r"
top: "pool2_r"
pooling_param {
pool: MAX
kernel_size: 3
stride: 2
}
}
layer {
name: "norm2_r"
type: "LRN"
bottom: "pool2_r"
top: "norm2_r"
lrn_param {
local_size: 5
alpha: 0.0001
beta: 0.75
}
}
layer {
name: "conv3_r"
type: "Convolution"
bottom: "norm2_r"
top: "conv3_r"
param {
name: "conv3_w"
lr_mult: 1
decay_mult: 1
}
param {
name: "conv3_b"
lr_mult: 2
decay_mult: 0
}
convolution_param {
num_output: 384
pad: 1
kernel_size: 3
}
}
layer {
name: "relu3_r"
type: "ReLU"
bottom: "conv3_r"
top: "conv3_r"
}
layer {
name: "conv4_r"
type: "Convolution"
bottom: "conv3_r"
top: "conv4_r"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 2
decay_mult: 0
}
convolution_param {
num_output: 64
pad: 0
kernel_size: 1
}
}
layer {
name: "relu4_r"
type: "ReLU"
bottom: "conv4_r"
top: "conv4_r"
}
layer {
name: "conv1_f"
type: "Convolution"
bottom: "image_face"
top: "conv1_f"
param {
name: "conv1_f_w"
lr_mult: 1
decay_mult: 1
}
param {
name: "conv1_f_b"
lr_mult: 2
decay_mult: 0
}
convolution_param {
num_output: 96
kernel_size: 11
stride: 4
}
}
layer {
name: "relu1_f"
type: "ReLU"
bottom: "conv1_f"
top: "conv1_f"
}
layer {
name: "pool1_f"
type: "Pooling"
bottom: "conv1_f"
top: "pool1_f"
pooling_param {
pool: MAX
kernel_size: 3
stride: 2
}
}
layer {
name: "norm1_f"
type: "LRN"
bottom: "pool1_f"
top: "norm1_f"
lrn_param {
local_size: 5
alpha: 0.0001
beta: 0.75
}
}
layer {
name: "conv2_f"
type: "Convolution"
bottom: "norm1_f"
top: "conv2_f"
param {
name: "conv2_f_w"
lr_mult: 1
decay_mult: 1
}
param {
name: "conv2_f_b"
lr_mult: 2
decay_mult: 0
}
convolution_param {
num_output: 256
pad: 2
kernel_size: 5
group: 2
}
}
layer {
name: "relu2_f"
type: "ReLU"
bottom: "conv2_f"
top: "conv2_f"
}
layer {
name: "pool2_f"
type: "Pooling"
bottom: "conv2_f"
top: "pool2_f"
pooling_param {
pool: MAX
kernel_size: 3
stride: 2
}
}
layer {
name: "norm2_f"
type: "LRN"
bottom: "pool2_f"
top: "norm2_f"
lrn_param {
local_size: 5
alpha: 0.0001
beta: 0.75
}
}
layer {
name: "conv3_f"
type: "Convolution"
bottom: "norm2_f"
top: "conv3_f"
param {
name: "conv3_f_w"
lr_mult: 1
decay_mult: 1
}
param {
name: "conv3_f_b"
lr_mult: 2
decay_mult: 0
}
convolution_param {
num_output: 384
pad: 1
kernel_size: 3
}
}
layer {
name: "relu3_f"
type: "ReLU"
bottom: "conv3_f"
top: "conv3_f"
}
layer {
name: "conv4_f"
type: "Convolution"
bottom: "conv3_f"
top: "conv4_f"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 2
decay_mult: 0
}
convolution_param {
num_output: 64
pad: 0
kernel_size: 1
}
}
layer {
name: "relu4_f"
type: "ReLU"
bottom: "conv4_f"
top: "conv4_f"
}
layer {
name: "fc1_f"
type: "InnerProduct"
bottom: "conv4_f"
top: "fc1_f"
param {
lr_mult: 1
}
param {
lr_mult: 2
}
inner_product_param {
num_output: 128
}
}
layer {
name: "relufc1_f"
type: "ReLU"
bottom: "fc1_f"
top: "fc1_f"
}
layer {
name: "fc2_f"
type: "InnerProduct"
bottom: "fc1_f"
top: "fc2_f"
param {
lr_mult: 1
}
param {
lr_mult: 2
}
inner_product_param {
num_output: 64
}
}
layer {
name: "relufc2_f"
type: "ReLU"
bottom: "fc2_f"
top: "fc2_f"
}
layer {
name: "concat1"
bottom: "conv4_l"
bottom: "conv4_r"
top: "concat1"
type: "Concat"
concat_param {
axis: 1
}
}
layer {
name: "fc1"
type: "InnerProduct"
bottom: "concat1"
top: "fc1"
param {
lr_mult: 1
}
param {
lr_mult: 2
}
inner_product_param {
num_output: 128
}
}
layer {
name: "relufc1"
type: "ReLU"
bottom: "fc1"
top: "fc1"
}
layer {
name: "flatten"
type: "Flatten"
bottom: "facegrid"
top: "flatten"
}
layer {
name: "fg_fc1"
type: "InnerProduct"
bottom: "flatten"
top: "fg_fc1"
param {
lr_mult: 1
}
param {
lr_mult: 2
}
inner_product_param {
num_output: 256
weight_filler {
type: "gaussian"
std: 0.005
}
bias_filler {
type: "constant"
value: 1
}
}
}
layer {
name: "relufgfc1"
type: "ReLU"
bottom: "fg_fc1"
top: "fg_fc1"
}
layer {
name: "fg_fc2"
type: "InnerProduct"
bottom: "fg_fc1"
top: "fg_fc2"
param {
lr_mult: 1
}
param {
lr_mult: 2
}
inner_product_param {
num_output: 128
}
}
layer {
name: "relufgfc2"
type: "ReLU"
bottom: "fg_fc2"
top: "fg_fc2"
}
layer {
name: "concat2"
bottom: "fc1"
bottom: "fg_fc2"
bottom: "fc2_f"
top: "concat2"
type: "Concat"
concat_param {
axis: 1
}
}
layer {
name: "fc2"
type: "InnerProduct"
bottom: "concat2"
top: "fc2"
param {
lr_mult: 1
}
param {
lr_mult: 2
}
inner_product_param {
num_output: 128
}
}
layer {
name: "relufc2"
type: "ReLU"
bottom: "fc2"
top: "fc2"
}
layer {
name: "fc3"
type: "InnerProduct"
bottom: "fc2"
top: "fc3"
param {
lr_mult: 1
}
param {
lr_mult: 2
}
inner_product_param {
num_output: 2
}
}
有四个输入，即：面部，右眼，左眼和面部网格。如果我单独进行，我可以将它们转换为四个cfg，但这将无法实现我所需要的。有什么线索吗？还是我必须手动编写一个cfg，我尝试了，但是却出现了预期的错误。欢迎输入。感谢您的时间。