forestOnFire.m

%% SVM主程序
clear;clc;
%% 对图片进行处理
% X存储图像数据，每一行为一幅图,不同的列代表不同的feature，Y为对应的0和1，1为onfire
% matlabpool local 4
tic
%processImg函数已经把X，Y数据保存在svmdata中了，直接load svmdata读取数据，不需要每次重新计算X和Y
% [X,Y]=processImg();
load svmdata
%% SVM using linear kernel

[C1, sigma1] = optParams(X,Y,'linearKernel');
[avgaccuracy1,avgoriginal1]=crossValidation(X,Y,7,'linearKernel',C1,sigma1);
fprintf('SVM model original data fitting precision using linear kernel: %2.2f%%\n',avgoriginal1*100);
fprintf('SVM model prediction precision using linear kernel: %2.2f%%\n',avgaccuracy1*100);

% figure(2)
% visualizeBoundaryLinear(X, Y, model);%size(X,1)=2, 2 feature

%% SVM using gaussian kernel
[C2, sigma2] = optParams(X,Y,'gaussianKernel');
[avgaccuracy2,avgoriginal2]=crossValidation(X,Y,7,'gaussianKernel',C2,sigma2);
fprintf('SVM model original data fitting precision using gaussian kernel: %2.2f%%\n',avgoriginal2*100);
fprintf('SVM model prediction precision using gaussian kernel: %2.2f%%\n',avgaccuracy2*100);

% figure(3)
% visualizeBoundary(X, Y, model);%size(X,1)=2, 2 feature


toc
% matlabpool close