caffe---mnist数据集训练与测试

1、数据。mnist_test_lmdb和mnist_train_lmdb数据

2、路径。

（1）修改lenet_train_test.prototxt文件，训练和测试两处

  source: "....省略/examples/mnist/mnist-train-leveldb" //写上你的绝对路径  backend: LEVELDB //格式改成LEVELDB

（2）修改lenet_solver.prototxt文件：

net: "....省略/examples/mnist/lenet_train_test.prototxt"  //绝对路径

snapshot_prefix: "....省略/examples/mnist/lenet" //绝对路径

solver_mode: CPU //CPU模式

---

3、右键caffe打开属性：

在Command Arguments输入： train --solver=前面的绝对路径/mnist/lenet_solver.prototxt

4、 确定后debug caffe,大功告成！

---

1、训练完后，会生成lenet_iter_5000.caffemodel，lenet_iter_5000.solverstate，lenet_iter_10000.caffemodel，lenet_iter_10000.solverstate四个文件

2、产生均值文件

计算均值文件：在E:\CaffeDev-GPU\caffe-master\Build\x64\Release目录下新建bat文件mnist_mean.bat，内容如下：

D:/caffewin/caffe-master/Build/x64/Debug/compute_image_mean.exe D:/caffewin/caffe-master/examples/mnist/mnist_train_lmdb mean.binaryproto --backend=lmdb
pause

双击运行

得到mean.binaryproto

3、新建mnist_test.bat

D:/caffewin/caffe-master/Build/x64/Debug/caffe.exe test --model=D:/caffewin/caffe-master/examples/mnist/lenet_train_test.prototxt --weights=D:/caffewin/caffe-master/examples/mnist/lenet_iter_10000.caffemodel
pause

双击运行

4、测试图片

新建test.bat文件，写入

D:\caffewin\caffe-master\Build\x64\Debug\classification.exe D:\caffewin\caffe-master\examples\mnist\lenet.prototxt D:\caffewin\caffe-master\examples\mnist\lenet_iter_10000.caffemodel D:\caffewin\caffe-master\examples\mnist\mean.binaryproto D:\caffewin\caffe-master\examples\mnist\words.txt D:\caffewin\caffe-master\examples\mnist\.bmp
pause  

双击运行即可