Clear Filters
Clear Filters

how to plot confusionmat for this code?

1 view (last 30 days)
susanta reang
susanta reang on 18 Jun 2018
close all clear all clc delete('Frames\*.jpg'); [filename pathname] = uigetfile({'*.avi'},'Select A Video File'); I = VideoReader([pathname,filename]); implay([pathname,filename]); pause(3); nFrames = I.numberofFrames; vidHeight = I.Height; vidWidth = I.Width; mov(1:nFrames) = ... struct('cdata', zeros(vidHeight, vidWidth, 3, 'uint8'),... 'colormap', []); WantedFrames = 50; for k = 1:WantedFrames mov(k).cdata = read( I, k); mov(k).cdata = imresize(mov(k).cdata,[256,256]); imwrite(mov(k).cdata,['Frames\',num2str(k),'.jpg']); end
for I = 1:WantedFrames im=imread(['Frames\',num2str(I),'.jpg']); figure(1),subplot(5,10,I),imshow(im); end clc for i=1:WantedFrames disp(['Processing frame no.',num2str(i)]); img=imread(['Frames\',num2str(i),'.jpg']); f1=il_rgb2gray(double(img)); [ysize,xsize]=size(f1); nptsmax=40; kparam=0.04; pointtype=1; sxl2=4; sxi2=2*sxl2; % detect points [posinit,valinit]=STIP(f1,kparam,sxl2,sxi2,pointtype,nptsmax); Test_Feat(i,1:40)=valinit; %imshow(f1,[]), hold on % axis off; % showellipticfeatures(posinit,[1 1 0]); % title('Feature Points','fontsize',12,'fontname','Times New Roman','color','Black') end
% Use KNN To classify the videos load('TrainFeature.mat') X = meas; Y = New_Label; Z = Test_Feat; % Now Classify
%ens = fitensemble(X,Y,'Subspace',300,'KNN'); %class = predict(ens,Z(1,:)) md1 = ClassificationKNN.fit(X,Y); Type = predict(md1,Z); if (Type == 1) disp('Boxing'); helpdlg(' Boxing '); elseif (Type == 2) disp('Hand Clapping'); helpdlg('Hand Clapping'); elseif (Type == 3) disp('Hand Waving'); helpdlg('Hand Waving'); elseif (Type == 4) disp('Jogging'); helpdlg('Jogging'); elseif (Type == 5) disp('Running'); helpdlg('Running'); elseif (Type == 6) disp('Walking'); helpdlg('Walking'); elseif (Type == 7) disp('Cycling'); helpdlg('Cycling'); elseif (Type == 8) disp('Surfing'); helpdlg('Surfing'); end

Answers (1)

Yuvaraj Venkataswamy
Yuvaraj Venkataswamy on 18 Jun 2018
if true
plotconfusion(actual_labels,Predicted_labels)
end
In this, Predicted_labels are which you have classified through KNN and actual_labels are the true labels.

Categories

Find more on Denoising and Compression in Help Center and File Exchange

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!