Convolutional neural network practical

A computer vision practical by the Oxford Visual Geometry group, authored by Andrea Vedaldi and Andrew Zisserman.

Start from doc/instructions.html.

Note that this practical requires compiling the (included) MatConvNet library. This should happen automatically (see the setup.m script), but make sure that the compilation succeeds on the laboratory computers.

Package contents

The practical consists of four exercises, organized in the following files:

  • exercise1.m -- Part 1: CNN fundamentals
  • exercise2.m -- Part 2: Derivatives and backpropagation
  • exercise3.m -- Part 3: Learning a tiny CNN
  • exercise4.m -- Part 4: Learning a CNN to recognize characters
  • exercise5.m -- Part 5: Using a pretrained CNN

The practical runs in MATLAB and uses MatConvNet and VLFeat. This package contains the following MATLAB functions:

  • extractBlackBlobs.m: extract black blobs from an image.
  • tinycnn.m: implements a very simple CNN.
  • initializeCharacterCNN.m: initialize a CNN to recognize characters.
  • decodeCharacters.m: visualize the output of the character CNN.
  • setup.m: setup MATLAB environment.

Appendix: Installing from scratch

The practical requires both VLFeat and MatConvNet. VLFeat comes with pre-built binaries, but MatConvNet does not.

  1. Set the current directory to the practical base directory.
  2. From Bash:
    1. Run ./extras/download.sh. This will download the imagenet-vgg-verydeep-16.mat model as well as a binary copy of the VLFeat library and a copy of MatConvNet.
    2. Run ./extra/genfonts.sh. This will download the Google Fonts and extract them as PNG files.
    3. Run ./extra/genstring.sh. This will create data/sentence-lato.png.
  3. From MATLAB run addpath extra ; packFonts ;. This will create data/charsdb.mat.
  4. Test the practical: from MATLAB run all the exercises in order.

Changes

  • 2015a - Initial edition

License

Copyright (c) 2015 Andrea Vedaldi

Permission is hereby granted, free of charge, to any person
obtaining a copy of this software and associated documentation
files (the "Software"), to deal in the Software without
restriction, including without limitation the rights to use, copy,
modify, merge, publish, distribute, sublicense, and/or sell copies
of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be
included in all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT
HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
DEALINGS IN THE SOFTWARE.


Convolutional neural network practical

牛津视觉几何组实用的计算机视觉, 由Andrea Vedaldi和Andrew Zisserman创作。

doc / instructions.html 开始。

Note that this practical requires compiling the (included) MatConvNet library. This should happen automatically (see the setup.m script), but make sure that the compilation succeeds on the laboratory computers.

包装内容

实践包括以下四个练习 档案:

  • exercise1.m - 第1部分:CNN基础知识
  • exercise2.m - 第2部分:衍生工具和反向传播
  • exercise3.m - 第3部分:学习一个微小的CNN
  • exercise4.m - 第4部分:学习CNN识别角色
  • exercise5.m - 第5部分:使用预先训练的CNN
实际运行在MATLAB和使用 MatConvNet VLFeat 。此包包含以下内容 MATLAB函数:

  • extractBlackBlobs.m :从图像中提取黑色斑点。
  • tinycnn.m :实现一个非常简单的CNN。
  • initializeCharacterCNN.m :初始化CNN以识别字符。
  • decodeCharacters.m :可视化CNN字符的输出。
  • setup.m :设置MATLAB环境。

附录:从头开始安装

实际需要VLFeat和MatConvNet。 VLFeat随附 预制的二进制文件,但是MatConvNet不支持。

  1. Set the current directory to the practical base directory.
  2. From Bash:
    1. Run ./extras/download.sh. This will download the imagenet-vgg-verydeep-16.mat model as well as a binary copy of the VLFeat library and a copy of MatConvNet.
    2. Run ./extra/genfonts.sh. This will download the Google Fonts and extract them as PNG files.
    3. Run ./extra/genstring.sh. This will create data/sentence-lato.png.
  3. From MATLAB run addpath extra ; packFonts ;. This will create data/charsdb.mat.
  4. Test the practical: from MATLAB run all the exercises in order.

更改

  • 2015a - 初始版

许可证

Copyright © 2015 Andrea Vedaldi

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.




相关问题推荐