Caffe部署中的几个train-test-solver-prototxt-deploy等说明二
Caffe部署中的幾個train-test-solver-prototxt-deploy等說明<二>
發表于2016/9/15 20:39:52 ?1049人閱讀
分類：?神經網絡與深度學習
一，train_val.prototxt
name: "CIFAR10_quick" layer {name: "cifar"type: "Data"top: "data"top: "label"include {phase: TRAIN}transform_param {# mirror: true# mean_file: "examples/cifar10/mean.binaryproto"uumean_file: "myself/00b/00bmean.binaryproto" }data_param {# source: "examples/cifar10/cifar10_train_lmdb"source: "myself/00b/00b_train_lmdb"batch_size: 50backend: LMDB} } layer {name: "cifar"type: "Data"top: "data"top: "label"include {phase: TEST}transform_param {# mean_file: "examples/cifar10/mean.binaryproto"mean_file: "myself/00b/00bmean.binaryproto"}data_param {# source: "examples/cifar10/cifar10_test_lmdb"source: "myself/00b/00b_val_lmdb"batch_size: 50backend: LMDB} } layer {name: "conv1"type: "Convolution"bottom: "data"top: "conv1"param {lr_mult: 1}param {lr_mult: 2}convolution_param {num_output: 32# pad: 1kernel_size: 4stride: 1weight_filler {type: "gaussian"std: 0.0001}bias_filler {type: "constant"}} } layer {name: "pool1"type: "Pooling"bottom: "conv1"top: "pool1"pooling_param {pool: MAXkernel_size: 2stride: 2} } layer {name: "relu1"type: "ReLU"bottom: "pool1"top: "pool1" } layer {name: "conv2"type: "Convolution"bottom: "pool1"top: "conv2"param {lr_mult: 1}param {lr_mult: 2}convolution_param {num_output: 32# pad: 2kernel_size: 4stride: 1weight_filler {type: "gaussian"std: 0.01}bias_filler {type: "constant"}} } layer {name: "relu2"type: "ReLU"bottom: "conv2"top: "conv2" } layer {name: "pool2"type: "Pooling"bottom: "conv2"top: "pool2"pooling_param {pool: AVEkernel_size: 2stride: 2} } layer {name: "conv3"type: "Convolution"bottom: "pool2"top: "conv3"param {lr_mult: 1}param {lr_mult: 2}convolution_param {num_output: 32# pad: 2kernel_size: 4stride: 1weight_filler {type: "gaussian"std: 0.01}bias_filler {type: "constant"}} } layer {name: "relu3"type: "ReLU"bottom: "conv3"top: "conv3" } layer {name: "pool3"type: "Pooling"bottom: "conv3"top: "pool3"pooling_param {pool: AVEkernel_size: 2stride: 2} } layer {name: "conv4"type: "Convolution"bottom: "pool3"top: "conv4"param {lr_mult: 1}param {lr_mult: 2}convolution_param {num_output: 32# pad: 2kernel_size: 4stride: 1weight_filler {type: "gaussian"std: 0.01}bias_filler {type: "constant"}} } layer {name: "relu4"type: "ReLU"bottom: "conv4"top: "conv4" } layer {name: "pool4"type: "Pooling"bottom: "conv4"top: "pool4"pooling_param {pool: AVEkernel_size: 2stride: 2} } layer {name: "ip1"type: "InnerProduct"bottom: "pool4"top: "ip1"param {lr_mult: 1}param {lr_mult: 2}inner_product_param {num_output: 200weight_filler {type: "gaussian"std: 0.1}bias_filler {type: "constant"}} } layer {name: "ip2"type: "InnerProduct"bottom: "ip1"top: "ip2"param {lr_mult: 1}param {lr_mult: 2}inner_product_param {num_output: 3weight_filler {type: "gaussian"std: 0.1}bias_filler {type: "constant"}} } layer {name: "accuracy"type: "Accuracy"bottom: "ip2"bottom: "label"top: "accuracy"include {phase: TEST} } layer {name: "loss"type: "SoftmaxWithLoss"bottom: "ip2"bottom: "label"top: "loss" }二，solver.prototxt
# reduce the learning rate after 8 epochs (4000 iters) by a factor of 10# The train/test net protocol buffer definition net: "myself/00b/train_val.prototxt" # test_iter specifies how many forward passes the test should carry out. # In the case of MNIST, we have test batch size 100 and 100 test iterations, # covering the full 10,000 testing images. test_iter: 10 # Carry out testing every 500 training iterations. test_interval: 70 # The base learning rate, momentum and the weight decay of the network. base_lr: 0.001 momentum: 0.9 weight_decay: 0.004 # The learning rate policy lr_policy: "fixed" # lr_policy: "step" gamma: 0.1 stepsize: 100 # Display every 100 iterations display: 10 # The maximum number of iterations max_iter: 2000 # snapshot intermediate results # snapshot: 3000 # snapshot_format: HDF5snapshot_prefix: "myself/00b/00b" # solver mode: CPU or GPU solver_mode: CPU三，deploy.prototxt
name: "CIFAR10_quick" layer {name: "data"type: "Input"top: "data"input_param { shape: { dim: 1 dim: 3 dim: 101 dim: 101 } } } layer {name: "conv1"type: "Convolution"bottom: "data"top: "conv1"convolution_param {num_output: 32kernel_size: 4stride: 1} } layer {name: "relu1"type: "ReLU"bottom: "conv1"top: "conv1" } layer {name: "pool1"type: "Pooling"bottom: "conv1"top: "pool1"pooling_param {pool: MAXkernel_size: 2stride: 2} } layer {name: "conv2"type: "Convolution"bottom: "pool1"top: "conv2"convolution_param {num_output: 32kernel_size: 4stride: 1} } layer {name: "relu2"type: "ReLU"bottom: "conv2"top: "conv2" } layer {name: "pool2"type: "Pooling"bottom: "conv2"top: "pool2"pooling_param {pool: MAXkernel_size: 2stride: 2} } layer {name: "conv3"type: "Convolution"bottom: "pool2"top: "conv3"convolution_param {num_output: 32kernel_size: 4stride: 1} } layer {name: "relu3"type: "ReLU"bottom: "conv3"top: "conv3" } layer {name: "pool3"type: "Pooling"bottom: "conv3"top: "pool3"pooling_param {pool: MAXkernel_size: 2stride: 2} } layer {name: "conv4"type: "Convolution"bottom: "pool3"top: "conv4"convolution_param {num_output: 32kernel_size: 4stride: 1} } layer {name: "relu4"type: "ReLU"bottom: "conv4"top: "conv4" } layer {name: "pool4"type: "Pooling"bottom: "conv4"top: "pool4"pooling_param {pool: MAXkernel_size: 2stride: 2} } layer {name: "ip1"type: "InnerProduct"bottom: "pool4"top: "ip1"inner_product_param {num_output: 200} } layer {name: "ip2"type: "InnerProduct"bottom: "ip1"top: "ip2"inner_product_param {num_output: 3} } layer {#name: "loss"name: "prob"type: "Softmax" bottom: "ip2"top: "prob"#top: "loss" }參考一：
模型就用程序自帶的caffenet模型，位置在?models/bvlc_reference_caffenet/文件夾下, 將需要的兩個配置文件，復制到myfile文件夾內
# sudo cp models/bvlc_reference_caffenet/solver.prototxt examples/myfile/ # sudo cp models/bvlc_reference_caffenet/train_val.prototxt examples/myfile/修改train_val.protxt，只需要修改兩個階段的data層就可以了，其它可以不用管。
name: "CaffeNet" layer {name: "data" type: "Data" top: "data" top: "label" include { phase: TRAIN } transform_param { mirror: true crop_size: 227 mean_file: "examples/myfile/mean.binaryproto" } data_param { source: "examples/myfile/img_train_lmdb" batch_size: 256 backend: LMDB } } layer { name: "data" type: "Data" top: "data" top: "label" include { phase: TEST } transform_param { mirror: false crop_size: 227 mean_file: "examples/myfile/mean.binaryproto" } data_param { source: "examples/myfile/img_test_lmdb" batch_size: 50 backend: LMDB } } ?實際上就是修改兩個data layer的mean_file和source這兩個地方，其它都沒有變化 。
修改其中的solver.prototxt
# sudo vi examples/myfile/solver.prototxt net: "examples/myfile/train_val.prototxt" test_iter: 2 test_interval: 50 base_lr: 0.001 lr_policy: "step" gamma: 0.1 stepsize: 100 display: 20 max_iter: 500 momentum: 0.9 weight_decay: 0.005 solver_mode: GPU100個測試數據，batch_size為50，因此test_iter設置為2，就能全cover了。在訓練過程中，調整學習率，逐步變小。
參考二：
前面做好了lmdb和均值文件，下面以Googlenet為例修改網絡并訓練模型。
?
我們將caffe-master\models下的bvlc_googlenet文件夾復制到caffe-master\examples\imagenet下。（因為我們的lmdb和均值都在這里，放一起方便些）
打開train_val.txt，修改：
1.修改data層：
?
2.修改輸出：
由于Googlenet有三個輸出，所以改三個地方，其他網絡一般只有一個輸出，則改一個地方即可。
如果是微調，那么輸出層的層名也要修改。（參數根據層名來初始化，由于輸出改了，該層參數就不對應了，因此要改名）
layer {name: "loss1/classifier"type: "InnerProduct"bottom: "loss1/fc"top: "loss1/classifier"param {lr_mult: 1decay_mult: 1}param {lr_mult: 2decay_mult: 0}inner_product_param {num_output: 1000 #改成你的數據集類別數weight_filler {type: "xavier"}bias_filler {type: "constant"value: 0}} } layer {name: "loss2/classifier"type: "InnerProduct"bottom: "loss2/fc"top: "loss2/classifier"param {lr_mult: 1decay_mult: 1}param {lr_mult: 2decay_mult: 0}inner_product_param {num_output: 1000 #改成你的數據集類別數weight_filler {type: "xavier"}bias_filler {type: "constant"value: 0}} } layer {name: "loss3/classifier"type: "InnerProduct"bottom: "pool5/7x7_s1"top: "loss3/classifier"param {lr_mult: 1decay_mult: 1}param {lr_mult: 2decay_mult: 0}inner_product_param {num_output: 1000 #改成你的數據集類別數weight_filler {type: "xavier"}bias_filler {type: "constant"value: 0}} }3.打開deploy.prototxt，修改：
layer {name: "loss3/classifier"type: "InnerProduct"bottom: "pool5/7x7_s1"top: "loss3/classifier"param {lr_mult: 1decay_mult: 1}param {lr_mult: 2decay_mult: 0}inner_product_param {num_output: 1000 #改成你的數據集類別數weight_filler {type: "xavier"}bias_filler {type: "constant"value: 0}} }如果是微調，該層層名和train_val.prototxt修改一致。
接著，打開solver，修改：
net: "examples/imagenet/bvlc_googlenet/train_val.prototxt" #路徑不要錯 test_iter: 1000 #前面已說明該值 test_interval: 4000 #迭代多少次測試一次 test_initialization: false display: 40 average_loss: 40 base_lr: 0.01 lr_policy: "step" stepsize: 320000 #迭代多少次改變一次學習率 gamma: 0.96 max_iter: 10000000 #迭代次數 momentum: 0.9 weight_decay: 0.0002 snapshot: 40000 snapshot_prefix: "examples/imagenet/bvlc_googlenet" #生成的caffemodel保存在imagenet下，形如bvlc_googlenet_iter_***.caffemodel solver_mode: GPU這時，我們回到caffe-master\examples\imagenet下，打開train_caffenet.sh，修改：
(如果是微調，在腳本里加入-weights **/**/**.caffemodel即可，即用來微調的caffemodel路徑)
[plain] #!/usr/bin/env sh./build/tools/caffe train \-solver examples/imagenet/bvlc_googlenet/solver.prototxt -gpu 0（如果有多個GPU，可自行選擇） 然后，在caffe-master下執行改腳本即可開始訓練:$caffe-master ./examples/imagenet/train_caffenet.sh
訓練得到的caffemodel就可以用來做圖像分類了，此時，需要（1）得到的labels.txt，（2）得到的mydata_mean.binaryproto，（3）得到的caffemodel以及已經修改過的deploy.prototxt，共四個文件，具體過程看：http://blog.csdn.net/sinat_30071459/article/details/50974695
參考三：
*_train_test.prototxt,*_deploy.prototxt,*_slover.prototxt文件編寫時注意
1、*_train_test.prototxt文件
這是訓練與測試網絡配置文件
（1）在數據層中 參數include{
???????????????????????????????? phase：TRAIN/TEST
?????????????????????????????}
TRAIN與TEST不能有“...”否則會報錯，還好提示信息里，會提示哪一行出現了問題，如下圖：
數字8就代表配置文件的第8行出現了錯誤
（2）卷積層和全連接層相似：卷積層（Convolution），全連接層（InnerProduct，容易翻譯成內積層）相似處有兩個【1】：都有兩個param{lr_mult：1
?????????????????????????????????????????? decay_mult：1????????????????????????????
?????????????????????????????? }
???????????????????????????? param{lr_mult： 2
?????????????????????????????????????? ?decay_mult: 0????????????
????????????????????????????? }
【2】：convolution_param{}與inner_product_param{}里面的參數相似，甚至相同
今天有事，明天再續！
續上！
（3）平均值文件*_mean.binaryproto要放在transform_param{}里，訓練與測試數據集放在data_param{}里
2.*_deploy.prototxt文件
【1】*_deploy.prototxt文件的構造和*_train_test.prototxt文件的構造稍有不同首先沒有test網絡中的test模塊，只有訓練模塊
【2】數據層的寫法和原來也有不同，更加簡潔：
input: "data" input_dim: 1 input_dim: 3 input_dim: 32 input_dim: 32
注意紅色部分，那是數據層的名字，沒有這個的話，第一卷積層無法找到數據，我一開始沒有加這句就報錯。下面的四個參數有點類似batch_size（1,3,32,32）里四個參數
【3】卷積層和全連接層中weight_filler{}與bias_filler{}兩個參數不用再填寫，應為這兩個參數的值，由已經訓練好的模型*.caffemodel文件提供
【4】輸出層的變化（1）沒有了test模塊測試精度（2）輸出層
*_train_test.prototxt文件：
layer{ ? name: "loss" ? type: "SoftmaxWithLoss"#注意此處與下面的不同 ? bottom: "ip2" ? bottom: "label"#注意標簽項在下面沒有了，因為下面的預測屬于哪個標簽，因此不能提供標簽 ? top: "loss" }
*_deploy.prototxt文件：
layer { ? name: "prob" ? type: "Softmax" ? bottom: "ip2" ? top: "prob" }
***注意在兩個文件中輸出層的類型都發生了變化一個是SoftmaxWithLoss，另一個是Softmax。另外為了方便區分訓練與應用輸出，訓練是輸出時是loss，應用時是prob。
3、*_slover.prototxt
net: "test.prototxt" #訓練網絡的配置文件 test_iter: 100 #test_iter 指明在測試階段有多上個前向過程（也就是有多少圖片）被執行。 在MNIST例子里，在網絡配置文件里已經設置test網絡的batch size=100，這里test_iter 設置為100，那在測試階段共有100*100=10000 圖片被處理 test_interval: 500 #每500次訓練迭代后，執行一次test base_lr: 0.01 #學習率初始化為0.01 momentum：0.9 #u=0.9 weight_decay:0.0005 # lr_policy: "inv" gamma: 0.0001 power: 0.75 #以上三個參數都和降低學習率有關，詳細的學習策略和計算公式見下面 // The learning rate decay policy. The currently implemented learning rate ?
// policies are as follows: ?
//??? - fixed: always return base_lr. ?
//??? - step: return base_lr * gamma ^ (floor(iter / step)) ?
//??? - exp: return base_lr * gamma ^ iter
//??? - inv: return base_lr * (1 + gamma * iter) ^ (- power) ?
//??? - multistep: similar to step but it allows non uniform steps defined by ?
//????? stepvalue ?
//??? - poly: the effective learning rate follows a polynomial decay, to be ?
//????? zero by the max_iter. return base_lr (1 - iter/max_iter) ^ (power) ?
//??? - sigmoid: the effective learning rate follows a sigmod decay ?
//????? return base_lr ( 1/(1 + exp(-gamma * (iter - stepsize)))) ?
// where base_lr, max_iter, gamma, step, stepvalue and power are defined ?
// in the solver parameter protocol buffer, and iter is the current iteration. display:100 #每100次迭代，顯示結果 snapshot: 5000 #每5000次迭代，保存一次快照 snapshot_prefix: "path_prefix" #快照保存前綴：更準確的說是快照保存路徑+前綴，應為文件名后的名字是固定的 solver_mode:GPU #選擇解算器是用cpu還是gpu
批處理文件編寫：
F:/caffe/caffe-windows-master/bin/caffe.exe train --solver=C:/Users/Administrator/Desktop/caffe_test/cifar-10/cifar10_slover_prototxt --gpu=all pause
參考四：
?
06?
原?
將train_val.prototxt 轉換成deploy.prototxt
?
分類：caffe?????????????????????????
（282）?????????????????????????????（0）?????????????????????????????????????????????????????????????????????????????????????????????????????????????????
?
1.刪除輸入數據（如：type:data...inckude{phase: TRAIN}），然后添加一個數據維度描述。
?
?
?
2.移除最后的“loss” 和“accuracy” 層，加入“prob”層。
?
?
?
總結
以上是生活随笔為你收集整理的Caffe部署中的几个train-test-solver-prototxt-deploy等说明二的全部內容，希望文章能夠幫你解決所遇到的問題。
- 上一篇： Caffe部署中的几个train-tes
- 下一篇： 4'.deploy.prototxt