caffe分类

1、制作数据  标签   label.bat

dir/s/on/b >0.txt

 

2、制作lmdb     convert_lmdb.bat

SET GLOG_logtostderr=1  
E:\caffeWin\caffe-master\Build\x64\Release\convert_imageset.exe --backend=lmdb --resize_width=32 --resize_height=32  --gray=false --shuffle G:\actionDataNew\ G:\actionDataNew\trainList.txt G:\actionDataNew\train_lmdb
E:\caffeWin\caffe-master\Build\x64\Release\convert_imageset.exe --backend=lmdb --resize_width=32 --resize_height=32  --gray=false --shuffle G:\actionDataNew\ G:\actionDataNew\valList.txt G:\actionDataNew\test_lmdb
pause  

 

3、计算均值  mean.bat

SET GLOG_logtostderr=1  
F:\CaffeVs2013\caffe-ssd-microsoft\Build\x64\Release\compute_image_mean.exe --backend=lmdb G:\actionDataNew\train_lmdb G:\actionDataNew\mean.binaryproto 
pause  

 

准备  train.prototxt 和 solver.prototxt

linux服务器训练

#!/usr/bin/env sh
set -e

TOOLS=/home/tsari/caffe/build/tools

$TOOLS/caffe train \
  --solver=/home/tsari/WorkSpace/OpenCloseFourLei/cifar10_quick_solver.prototxt $@

 

转载于:https://www.cnblogs.com/crazybird123/p/10338658.html