caffe 利用VGG训练自己的数据
写这个是因为有童鞋在跑VGG的时候遇到各种问题，供参考一下。
网络结构
以VGG16为例，自己跑的细胞数据
solver.prototxt：
net: "/media/dl/source/Experiment/cell/test/vgg/vgg16.prototxt"
test_iter:
test_interval:
base_lr: 0.0001
lr_policy: "step"
gamma: 0.1
stepsize:
display:
max_iter:
momentum: 0.9
weight_decay: 0.0005
snapshot:
snapshot_prefix: "/media/dl/source/Experiment/cell/test/vgg/vgg"
solver_mode: GPU
vgg16.prototxt：
注意，这里的数据层我是用的“ImageData”格式，也就是没有转为LMDB，直接导入图片进去的，因为我用的服务器，为了方便。如果为了更高效，还是使用LMDB数据库的形式。使用LMDB数据库形式的数据层我也写了下，放在这个prototxt后面作为补充。
另外，注意修改最后一个全连接层的num_output为自己的类别数。并修改该层的名字，如我改为了“cellfc8”，是为了finetune vgg时重新训练该层，不使用该层的预训练参数。
name: "VGG16"
layer {
name: "data"
type: "ImageData"
top: "data"
top: "label"
include {
phase: TRAIN
}
# transform_param {
# mirror: true
# crop_size:
# mean_file: "data/ilsvrc12_shrt_256/imagenet_mean.binaryproto"
# } image_data_param {
source: "/media/dl/source/Experiment/cell/data/trainnew2_resize/trainnew.txt"
batch_size:
shuffle:true
#is_color: false
new_height:
new_width:
}
}
layer {
name: "data"
type: "ImageData"
top: "data"
top: "label"
include {
phase: TEST
}
# transform_param {
# mirror: false
# crop_size:
# mean_file: "data/ilsvrc12_shrt_256/imagenet_mean.binaryproto"
# } image_data_param {
source: "/media/dl/source/Experiment/cell/data/val2_resize/valnew.txt"
batch_size:
#is_color: false
new_height:
new_width:
}
}
layer {
bottom: "data"
top: "conv1_1"
name: "conv1_1"
type: "Convolution"
param {
lr_mult:
decay_mult:
}
param {
lr_mult:
decay_mult:
}
convolution_param {
num_output:
pad:
kernel_size:
weight_filler {
type: "gaussian"
std: 0.01
}
bias_filler {
type: "constant"
value:
}
}
}
layer {
bottom: "conv1_1"
top: "conv1_1"
name: "relu1_1"
type: "ReLU"
}
layer {
bottom: "conv1_1"
top: "conv1_2"
name: "conv1_2"
type: "Convolution"
param {
lr_mult:
decay_mult:
}
param {
lr_mult:
decay_mult:
}
convolution_param {
num_output:
pad:
kernel_size:
weight_filler {
type: "gaussian"
std: 0.01
}
bias_filler {
type: "constant"
value:
}
}
}
layer {
bottom: "conv1_2"
top: "conv1_2"
name: "relu1_2"
type: "ReLU"
}
layer {
bottom: "conv1_2"
top: "pool1"
name: "pool1"
type: "Pooling"
pooling_param {
pool: MAX
kernel_size:
stride:
}
}
layer {
bottom: "pool1"
top: "conv2_1"
name: "conv2_1"
type: "Convolution"
param {
lr_mult:
decay_mult:
}
param {
lr_mult:
decay_mult:
}
convolution_param {
num_output:
pad:
kernel_size:
weight_filler {
type: "gaussian"
std: 0.01
}
bias_filler {
type: "constant"
value:
}
}
}
layer {
bottom: "conv2_1"
top: "conv2_1"
name: "relu2_1"
type: "ReLU"
}
layer {
bottom: "conv2_1"
top: "conv2_2"
name: "conv2_2"
type: "Convolution"
param {
lr_mult:
decay_mult:
}
param {
lr_mult:
decay_mult:
}
convolution_param {
num_output:
pad:
kernel_size:
weight_filler {
type: "gaussian"
std: 0.01
}
bias_filler {
type: "constant"
value:
}
}
}
layer {
bottom: "conv2_2"
top: "conv2_2"
name: "relu2_2"
type: "ReLU"
}
layer {
bottom: "conv2_2"
top: "pool2"
name: "pool2"
type: "Pooling"
pooling_param {
pool: MAX
kernel_size:
stride:
}
}
layer {
bottom: "pool2"
top: "conv3_1"
name: "conv3_1"
type: "Convolution"
param {
lr_mult:
decay_mult:
}
param {
lr_mult:
decay_mult:
}
convolution_param {
num_output:
pad:
kernel_size:
weight_filler {
type: "gaussian"
std: 0.01
}
bias_filler {
type: "constant"
value:
}
}
}
layer {
bottom: "conv3_1"
top: "conv3_1"
name: "relu3_1"
type: "ReLU"
}
layer {
bottom: "conv3_1"
top: "conv3_2"
name: "conv3_2"
type: "Convolution"
param {
lr_mult:
decay_mult:
}
param {
lr_mult:
decay_mult:
}
convolution_param {
num_output:
pad:
kernel_size:
weight_filler {
type: "gaussian"
std: 0.01
}
bias_filler {
type: "constant"
value:
}
}
}
layer {
bottom: "conv3_2"
top: "conv3_2"
name: "relu3_2"
type: "ReLU"
}
layer {
bottom: "conv3_2"
top: "conv3_3"
name: "conv3_3"
type: "Convolution"
param {
lr_mult:
decay_mult:
}
param {
lr_mult:
decay_mult:
}
convolution_param {
num_output:
pad:
kernel_size:
weight_filler {
type: "gaussian"
std: 0.01
}
bias_filler {
type: "constant"
value:
}
}
}
layer {
bottom: "conv3_3"
top: "conv3_3"
name: "relu3_3"
type: "ReLU"
}
layer {
bottom: "conv3_3"
top: "pool3"
name: "pool3"
type: "Pooling"
pooling_param {
pool: MAX
kernel_size:
stride:
}
}
layer {
bottom: "pool3"
top: "conv4_1"
name: "conv4_1"
type: "Convolution"
param {
lr_mult:
decay_mult:
}
param {
lr_mult:
decay_mult:
}
convolution_param {
num_output:
pad:
kernel_size:
weight_filler {
type: "gaussian"
std: 0.01
}
bias_filler {
type: "constant"
value:
}
}
}
layer {
bottom: "conv4_1"
top: "conv4_1"
name: "relu4_1"
type: "ReLU"
}
layer {
bottom: "conv4_1"
top: "conv4_2"
name: "conv4_2"
type: "Convolution"
param {
lr_mult:
decay_mult:
}
param {
lr_mult:
decay_mult:
}
convolution_param {
num_output:
pad:
kernel_size:
weight_filler {
type: "gaussian"
std: 0.01
}
bias_filler {
type: "constant"
value:
}
}
}
layer {
bottom: "conv4_2"
top: "conv4_2"
name: "relu4_2"
type: "ReLU"
}
layer {
bottom: "conv4_2"
top: "conv4_3"
name: "conv4_3"
type: "Convolution"
param {
lr_mult:
decay_mult:
}
param {
lr_mult:
decay_mult:
}
convolution_param {
num_output:
pad:
kernel_size:
weight_filler {
type: "gaussian"
std: 0.01
}
bias_filler {
type: "constant"
value:
}
}
}
layer {
bottom: "conv4_3"
top: "conv4_3"
name: "relu4_3"
type: "ReLU"
}
layer {
bottom: "conv4_3"
top: "pool4"
name: "pool4"
type: "Pooling"
pooling_param {
pool: MAX
kernel_size:
stride:
}
}
layer {
bottom: "pool4"
top: "conv5_1"
name: "conv5_1"
type: "Convolution"
param {
lr_mult:
decay_mult:
}
param {
lr_mult:
decay_mult:
}
convolution_param {
num_output:
pad:
kernel_size:
weight_filler {
type: "gaussian"
std: 0.01
}
bias_filler {
type: "constant"
value:
}
}
}
layer {
bottom: "conv5_1"
top: "conv5_1"
name: "relu5_1"
type: "ReLU"
}
layer {
bottom: "conv5_1"
top: "conv5_2"
name: "conv5_2"
type: "Convolution"
param {
lr_mult:
decay_mult:
}
param {
lr_mult:
decay_mult:
}
convolution_param {
num_output:
pad:
kernel_size:
weight_filler {
type: "gaussian"
std: 0.01
}
bias_filler {
type: "constant"
value:
}
}
}
layer {
bottom: "conv5_2"
top: "conv5_2"
name: "relu5_2"
type: "ReLU"
}
layer {
bottom: "conv5_2"
top: "conv5_3"
name: "conv5_3"
type: "Convolution"
param {
lr_mult:
decay_mult:
}
param {
lr_mult:
decay_mult:
}
convolution_param {
num_output:
pad:
kernel_size:
weight_filler {
type: "gaussian"
std: 0.01
}
bias_filler {
type: "constant"
value:
}
}
}
layer {
bottom: "conv5_3"
top: "conv5_3"
name: "relu5_3"
type: "ReLU"
}
layer {
bottom: "conv5_3"
top: "pool5"
name: "pool5"
type: "Pooling"
pooling_param {
pool: MAX
kernel_size:
stride:
}
}
layer {
bottom: "pool5"
top: "fc6"
name: "fc6"
type: "InnerProduct"
param {
lr_mult:
decay_mult:
}
param {
lr_mult:
decay_mult:
}
inner_product_param {
num_output:
weight_filler {
type: "gaussian"
std: 0.005
}
bias_filler {
type: "constant"
value: 0.1
}
}
}
layer {
bottom: "fc6"
top: "fc6"
name: "relu6"
type: "ReLU"
}
layer {
bottom: "fc6"
top: "fc6"
name: "drop6"
type: "Dropout"
dropout_param {
dropout_ratio: 0.5
}
}
layer {
bottom: "fc6"
top: "fc7"
name: "fc7"
type: "InnerProduct"
param {
lr_mult:
decay_mult:
}
param {
lr_mult:
decay_mult:
}
inner_product_param {
num_output:
weight_filler {
type: "gaussian"
std: 0.005
}
bias_filler {
type: "constant"
value: 0.1
}
}
}
layer {
bottom: "fc7"
top: "fc7"
name: "relu7"
type: "ReLU"
}
layer {
bottom: "fc7"
top: "fc7"
name: "drop7"
type: "Dropout"
dropout_param {
dropout_ratio: 0.5
}
}
layer {
bottom: "fc7"
top: "fc8"
name: "cellfc8"
type: "InnerProduct"
param {
lr_mult:
decay_mult:
}
param {
lr_mult:
decay_mult:
}
inner_product_param {
num_output: 7 #改为自己的类别数
weight_filler {
type: "gaussian"
std: 0.005
}
bias_filler {
type: "constant"
value: 0.1
}
}
}
layer {
name: "accuracy_at_1"
type: "Accuracy"
bottom: "fc8"
bottom: "label"
top: "accuracy_at_1"
accuracy_param {
top_k:
}
include {
phase: TEST
}
}
layer {
name: "accuracy_at_5"
type: "Accuracy"
bottom: "fc8"
bottom: "label"
top: "accuracy_at_5"
accuracy_param {
top_k:
}
include {
phase: TEST
}
}
layer {
bottom: "fc8"
bottom: "label"
top: "loss"
name: "loss"
type: "SoftmaxWithLoss"
}
如果使用LMDB数据库形式，将前面的数据层改为：
name: "vgg"
layer {
name: "data"
type: "Data"
top: "data"
top: "label"
include {
phase: TRAIN
}
transform_param {
mirror: true
crop_size:
#如果图片大于224，则使用crop的方式，小于则使用下面的new_height和new_width
# new_height:
#new_width:
mean_file: "vggface/face_mean.binaryproto"
}
data_param {
source: "vggface/face_train_lmdb"
batch_size:
backend: LMDB
}
}
layer {
name: "data"
type: "Data"
top: "data"
top: "label"
include {
phase: TEST
}
transform_param {
mirror: false
crop_size:
#如果图片大于224，则使用crop的方式，小于则使用下面的new_height和new_width
# new_height:
#new_width:
mean_file: "vggface/face_mean.binaryproto"
}
data_param {
source: "vggface/face_val_lmdb"
batch_size:
backend: LMDB
}
}
训练
放一个shell命令：
#!/usr/bin/env sh TOOLS=/home/dl/caffe-jonlong/build/tools $TOOLS/caffe train \
-solver=/media/dl/source/Experiment/cell/test/vgg/solver.prototxt \
-weights=/media/dl/source/Experiment/cell/test/vgg/VGG_ILSVRC_16_layers.caffemodel \
-gpu=all \
预训练模型VGG_ILSVRC_16_layers.caffemodel的下载地址为
caffe 利用VGG训练自己的数据的更多相关文章
- 利用YOLOV3训练自己的数据
写在前面:YOLOV3只有修改了源码才需要重新make,而且make之前要先make clean. 一.准备数据 在/darknet/VOCdevkit1下建立文件夹VOC2007. voc2007文 ...
- caffe 如何训练自己的数据图片
申明:此教程加工于caffe 如何训练自己的数据图片 一.准备数据 有条件的同学,可以去imagenet的官网http://www.image-net.org/download-images,下载im ...
- caffe学习三：使用Faster RCNN训练自己的数据
本文假设你已经完成了安装,并可以运行demo.py 不会安装且用PASCAL VOC数据集的请看另来两篇博客. caffe学习一:ubuntu16.04下跑Faster R-CNN demo (基于c ...
- caffe 用faster rcnn 训练自己的数据 遇到的问题
1 . 怎么处理那些pyx和.c .h文件 在lib下有一些文件为.pyx文件,遇到不能import可以cython 那个文件,然后把lib文件夹重新make一下. 遇到.c 和 .h一样的操作. 2 ...
- YOLO2解读，训练自己的数据及相关转载以供学习
https://pjreddie.com/darknet/yolo/ 具体安装及使用可以参考官方文档https://github.com/pjreddie/darknet https://blog.c ...
- YOLOv3：训练自己的数据（附优化与问题总结）
环境说明 系统:ubuntu16.04 显卡:Tesla k80 12G显存 python环境: 2.7 && 3.6 前提条件:cuda9.0 cudnn7.0 opencv3.4. ...
- 人脸检测及识别python实现系列（3）——为模型训练准备人脸数据
人脸检测及识别python实现系列(3)——为模型训练准备人脸数据 机器学习最本质的地方就是基于海量数据统计的学习,说白了,机器学习其实就是在模拟人类儿童的学习行为.举一个简单的例子,成年人并没有主动 ...
- TensorFlow下利用MNIST训练模型识别手写数字
本文将参考TensorFlow中文社区官方文档使用mnist数据集训练一个多层卷积神经网络(LeNet5网络),并利用所训练的模型识别自己手写数字. 训练MNIST数据集,并保存训练模型 # Pyth ...
- py-faster-rcnn 训练自己的数据
转载:http://blog.csdn.net/sinat_30071459/article/details/51332084 Faster-RCNN+ZF用自己的数据集训练模型(Python版本) ...
随机推荐
- 【vim】自动补全 Ctrl+n
Vim 默认有自动补全的功能.的确这个功能是很基本的,并且可以通过插件来增强,但它也很有帮助.方法很简单. Vim 尝试通过已经输入的单词来预测单词的结尾. 比如当你在同一个文件中第二次输入 &quo ...
- ARMV8 datasheet学习笔记3：AArch64应用级体系结构之Memory Type and Attributes
1.前言 2. Memory类型和属性 memory分为normal memory和device memory,两种类型的Memory有各自的属性,除了下面介绍的几种属性外,还有其他一些杂项属性 2. ...
- Caching漫谈--关于Cache的几个理论
如今缓存是随处可见了,如果你的程序还没有使用到缓存,那可能是你的程序并发量很低,或对实时性要求很低.我们公司的ERP在显示某些报表时,每次打开都需要花上几分钟的时间,假如搜索引擎也是这么慢,我想这家搜 ...
- mysql系列八、mysql数据库优化、慢查询优化、执行计划分析
mysql的性能优化无法一蹴而就,必须一步一步慢慢来,从各个方面进行优化,最终性能就会有大的提升. 一.介绍 对mysql优化是一个综合性的技术,主要包括 表的设计合理化(符合3NF) 添加适当索引( ...
- mysql当查询某字段结果为空并赋值
1 代码 1.1 当当前字段为空,查询结果返回“none”,并且统计出现频率 select case when 字段 is null then 'none' else 字段 end as 字段, co ...
- TCP 远程执行CMD (解决粘包问题) 代码
服务端 from socket import * import subprocess,json,struct server= socket(AF_INET,SOCK_STREAM) server.bi ...
- 大数据处理算法--Bloom Filter布隆过滤
1. Bloom-Filter算法简介 Bloom-Filter,即布隆过滤器,1970年由Bloom中提出.它可以用于检索一个元素是否在一个集合中. Bloom Filter(BF)是一种空间效率很 ...
- SeaJS入门教程系列之使用SeaJS（二）
SeaJS入门教程系列之使用SeaJS(二) 作者: 字体:[增加 减小] 类型:转载 时间:2014-03-03我要评论 这篇文章主要介绍了SeaJS入门教程系列之使用SeaJS,着重介绍了SeaJ ...
- JS定义一个立即执行的可重用函数
我定义了一个函数表达式 testFun var testFun = (function() { ... //函数内容})(); 测试结果:虽然 testFun 函数有如愿在页面加载后立即被执行,但再次 ...
- python+selenium八：Alert弹窗
此弹窗是浏览器自带的弹窗,不是html中的元素 from selenium import webdriverfrom selenium.webdriver.common.action_chains i ...