天天向上

conv-net: 下載

conv-net-0.1-prealpha.tar解壓縮至 C:\20150202wafer\cnet\conv-net-0.1

新增Win32主控台應用程式

專案名稱: sample

<加入><現有項目>

加入現有項目 *.c

點選<標頭檔><加入><現有項目>

加入現有項目 *.h

Include目錄加入

C:\OpenCV-2.4.9\opencv\build\include\

C:\OpenCV-2.4.9\opencv\build\include\opencv

設定其他程式庫目錄

C:\opencv-2.4.9\lib

 

加入下列全部*.lib

opencv_calib3d249.lib
opencv_contrib249.lib
opencv_core249.lib
opencv_features2d249.lib
opencv_flann249.lib
opencv_gpu249.lib
opencv_highgui249.lib
opencv_imgproc249.lib
opencv_legacy249.lib
opencv_ml249.lib
opencv_nonfree249.lib
opencv_objdetect249.lib
opencv_ocl249.lib
opencv_photo249.lib
opencv_stitching249.lib
opencv_superres249.lib
opencv_ts249.lib
opencv_video249.lib
opencv_videostab249.lib

在此, sample專案先告一段落, 準備安裝sample專案中所需的expat函式庫

----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

The Expat XML Parser 下載及安裝

下載後執行 expat-win32bin-2.1.0.exe

安裝完成後可以開啟路徑 C:\Program Files (x86)\Expat 2.1.0\

 

開啟方案 expat.dsw

編譯expat專案

按下編譯, 成功應該會跳出下列對話框

C:\Program Files (x86)\Expat 2.1.0\Source\win32\bin\Release

libexpat函式庫產生成功, 離成功的路又更近些!

----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

接著, 回到原本的sample專案

Include目錄加入

C:\Program Files (x86)\Expat 2.1.0\Source\lib                 <--------- expat函式庫

C:\20150202wafer\cnet\conv-net-0.1\include                    <--------- convolutional neural network

其他程式庫目錄

C:\Program Files (x86)\Expat 2.1.0\Source\win32\bin\Release

<輸入><其他相依性>加入libexpat.lib

編譯sample專案, 跳出錯誤訊息 libexpat.dll 對話框

複製來源目錄:   C:\Program Files (x86)\Expat 2.1.0\Source\win32\bin\Release

下的libexpat.dll

貼到目的目錄:   C:\20150202wafer\cnet\sample\Release

-------------------------------------------------------------------------------------------------------------------------------------------------------------------------

複製來源目錄:  C:\20150202wafer\cnet\conv-net-0.1\data

下的全部檔案

貼到目的目錄:   C:\20150202wafer\cnet\sample\Release

測試辨識

> sample mnist.xml testimg-1.png

 

相關資料:

ConvNet - C++ library for convolutional neural networks    <------- 只有測試訓練好的CNN

Neural Network for Recognition of Handwritten Digits  <------- 理論背景說明, 提供訓練及測試階段並包含原始程式碼MFC                  

myCNN  <------- Google工程師自行改寫成MATLAB版本