Index in position 2 exceeds array bounds (must not exceed 68160).

Hello all, I trying to run a program, but I get this error. Could you please help me? THANKS!!!!

This is the code

demoSBD.m
% This scripts train and test CHM on Stanford Background Dataset.
% Change the trainpath, labelpath to point to the SBD.
% I have a bash script, trainparallel.sh, to run CHM for 8 classes in
% paralle using trainparalle.m. If you have enough resources you can use that script instead of
% this one.
% I also have testparallel.sh and testparallel.m scripts to run testing in parallel. 

trainpath = '/home/smhf/Documents/matlab/code/iccv09Data/images/'; %path to the training Images
labelpath = '/home/smhf/Documents/matlab/code/iccv09Data/labels/'; %path to the training Labels

addpath(genpath('FilterStuff')); %add path to functions required for feature extraction.
% parameters
Nstage = 2;
Nlevel = 5;
load CV1.mat % This is the list of training and testing for the first cross validation.

% Only for preallocation purpose
filestr = dir([trainpath '*.jpg']);
filesla = dir([labelpath '*.txt']);
filestr = filestr(rTrain);
filesla = filesla(rTrain);
ntr = length(filestr);

for class = 0:7
    savingpath = ['./temp/' num2str(class) '/']; %path to save results and temporary files
    mkdir(savingpath);
    PixN = zeros(Nlevel+1,1);
    Nfeat = zeros(Nlevel+1,1);
    for l = 0:Nlevel
        for i = 1:ntr
            temp = load([labelpath filesla(i).name]);
            [M,~,indSu] = extractBinary(temp,class,l-1); 
            PixN(l+1) = PixN(l+1) + numel(indSu);
        end
    end
    
    for l = 0:Nlevel
        tempfeat = Filterbank(imread([trainpath filestr(1).name]),l);
        Nfeat(l+1) = size(tempfeat,1);
    end
    tempim = imread([trainpath filestr(1).name]);
    tempfeat = ConstructNeighborhoodsS(tempim(:,:,1));
    Nfeatcontext = size(tempfeat,1);

    param.ntr = ntr;
    param.PixN = PixN;
    param.Nfeat = Nfeat;
    param.Nfeatcontext = Nfeatcontext;
    param.Nlevel = Nlevel;
    param.class = class;
    param.rTrain = rTrain;


    % Train the CHM
    for s = 1:Nstage
        param.stage = s;
        for l = 0:Nlevel
            param.level = l;
            param.PixN = PixN(l+1);
            param.Nfeat = Nfeat(l+1);
            model = trainCHM(trainpath,labelpath,savingpath,param);
            if s==Nstage, break; end
        end
    end
end

% Test CHM
for class = 0:7
    savingpath = ['./temp/' num2str(class) '/'];
    param.Nstage = Nstage;
    testpath = './training/'; %path to the test Images
    fileste = dir([testpath '*.jpg']);
    fileste = fileste(rTest);
    str = [savingpath 'output_testImages/'];
    mkdir(str);
    outputs{length(fileste)} = [];
    parfor i = 1:length(fileste)
        img = imread([testpath fileste(i).name]);
        clabels = testCHM(img,savingpath,param);
        outputs{i} = clabels;
    end

    for i = 1:length(fileste)
     clabels = outputs{i};
     save([str 'slice' num2str(i)],'clabels','-v7.3');
    end
end

trainCHM.m
function model = trainCHM(trainpath,labelpath,savingpath,param)


ntr = param.ntr;
PixN = param.PixN;
Nfeat = param.Nfeat;
level = param.level;
stage = param.stage;
Nfeatcontext = param.Nfeatcontext;
Nlevel = param.Nlevel;
class = param.class;
rTrain = param.rTrain;

filestr = dir([trainpath '*.jpg']);
filesla = dir([labelpath '*.txt']);
filestr = filestr(rTrain);
filesla = filesla(rTrain);

if ~exist([savingpath 'MODEL_level' num2str(level) '_stage' num2str(stage) '.mat'],'file')
    % Feature Extraction
    fprintf('Extracting features ... stage %d level %d \n',stage,level);

    if stage==1 || level~=0
        X = zeros(Nfeat + level*Nfeatcontext,PixN);
        DD = zeros(1,PixN);
        COUNTER = 0;
        for i = 1:ntr
            img = imread([trainpath filestr(i).name]);
            img = MyDownSample(img,level-1);
            fv = Filterbank(img,level);

            fvcontext = zeros(level*Nfeatcontext,size(img,1)*size(img,2));
            for j = 0:level-1
                Contextpath = [savingpath 'output_level' num2str(j) '_stage' num2str(stage) '/'];
                temp = load([Contextpath 'slice' num2str(i)]);
                if j>=1
                    temp = MyDownSample(temp.clabels,level-j);
                elseif j==0
                    temp = MyDownSample(temp.clabels,level-j-1);
                end
                fvcontext(j*Nfeatcontext+1:(j+1)*Nfeatcontext,:) = ConstructNeighborhoodsS(temp);
            end

            temp = load([labelpath filesla(i).name]);
            [clabels,~,indSu] = extractBinary(temp,class,level-1);
            D = reshape(clabels,1,numel(clabels));
            D = D(indSu);
            fv = fv(:,indSu);
            fvcontext = fvcontext(:,indSu);

            X(:,COUNTER+1:COUNTER+size(fv,2)) = [fv;fvcontext];
            DD(1,COUNTER+1:COUNTER+size(fv,2)) = D;
            COUNTER = COUNTER + size(fv,2);
        end
    else
        X = zeros(Nfeat + (Nlevel+1)*Nfeatcontext,PixN);
        DD = zeros(1,PixN);
        COUNTER = 0;
        for i = 1:ntr
            img = imread([trainpath filestr(i).name]);
            img = MyDownSample(img,level-1);
            fv = Filterbank(img,level);

            fvcontext = zeros((Nlevel+1)*Nfeatcontext,size(img,1)*size(img,2));
            for j = 0:Nlevel
                Contextpath = [savingpath 'output_level' num2str(j) '_stage' num2str(stage-1) '/'];
                temp = load([Contextpath 'slice' num2str(i)]);
                temp = MyUpSample(temp.clabels,j-1);
                temp = temp(1:size(img,1),1:size(img,2),:);
                fvcontext(j*Nfeatcontext+1:(j+1)*Nfeatcontext,:) = ConstructNeighborhoodsS(temp);
            end


            temp = load([labelpath filesla(i).name]);
            [clabels,~,indSu] = extractBinary(temp,class,level-1);
            D = reshape(clabels,1,numel(clabels));
            D = D(indSu);
            fv = fv(:,indSu);
            fvcontext = fvcontext(:,indSu);

            X(:,COUNTER+1:COUNTER+size(fv,2)) = [fv;fvcontext];
            DD(1,COUNTER+1:COUNTER+size(fv,2)) = D;
            COUNTER = COUNTER + size(fv,2);
        end
    end
    % Subsampling
%     if PixN > 2000000 % incrrease this for real problems
        ind = [];
        maxs =min([histc(DD(:),0),histc(DD(:),1),1500000]);
        for i=0:max(DD)
            if histc(DD,i)>maxs
                a = find(DD==i);
                rpm = randperm(numel(a));
                index = a(rpm(1:maxs));
            else
                index = find(DD==i);
            end
            ind = [ind index];
        end
        X = X(:,ind);
        DD = DD(:,ind);
%     end
    % Learning the Classifier
    fprintf('start learning LDNN ... stage %d level %d \n',stage,level);
    opt.level = level;
    opt.stage = stage;
    tic;model = LearnAndOrNetMEX( X,DD,opt);Time = toc;
    save([savingpath 'MODEL_level' num2str(level) '_stage' num2str(stage)] , 'model','Time', '-v7.3');
else
    fprintf('Model_level %d _stage %d exists...proceed to generate output \n',level,stage);
    load([savingpath 'MODEL_level' num2str(level) '_stage' num2str(stage) '.mat']);
end
    % Write the outputs
fprintf('Generating outputs ... stage %d level %d \n',stage,level);
str = [savingpath 'output_level' num2str(level) '_stage' num2str(stage) '/'];
mkdir(str);

if ~exist([str 'slice' num2str(ntr) '.mat'],'file')
% matlabpool open 12
    trOutput{ntr} = [];
    if stage==1 || level~=0
        parfor i = 1:ntr
            img = imread([trainpath filestr(i).name]);
            img = MyDownSample(img,level-1);
            fv = Filterbank(img,level);

            fvcontext = zeros(level*Nfeatcontext,size(img,1)*size(img,2));
            for j = 0:level-1
                Contextpath = [savingpath 'output_level' num2str(j) '_stage' num2str(stage) '/'];
                temp = load([Contextpath 'slice' num2str(i)]);
                if j>=1
                    temp = MyDownSample(temp.clabels,level-j);
                elseif j==0
                    temp = MyDownSample(temp.clabels,level-j-1);
                end
                fvcontext(j*Nfeatcontext+1:(j+1)*Nfeatcontext,:) = ConstructNeighborhoodsS(temp);
            end

            X = [fv;fvcontext];
            [~, Y_floats] = EvaluateAndOrNetMX(X,model);
            clabels = reshape(Y_floats,[size(img,1) size(img,2)]);
            trOutput{i} = clabels;
        end
    else
        parfor i = 1:ntr
            img = imread([trainpath filestr(i).name]);
            img = MyDownSample(img,level-1);
            fv = Filterbank(img,level);

            fvcontext = zeros((Nlevel+1)*Nfeatcontext,size(img,1)*size(img,2));
            for j = 0:Nlevel
                Contextpath = [savingpath 'output_level' num2str(j) '_stage' num2str(stage-1) '/'];
                temp = load([Contextpath 'slice' num2str(i)]);
                temp = MyUpSample(temp.clabels,j-1);
                temp = temp(1:size(img,1),1:size(img,2),:);
                fvcontext(j*Nfeatcontext+1:(j+1)*Nfeatcontext,:) = ConstructNeighborhoodsS(temp);
            end

            X = [fv;fvcontext];
            [~, Y_floats] = EvaluateAndOrNetMX(X,model);
            clabels = reshape(Y_floats,[size(img,1) size(img,2)]);
            trOutput{i} = clabels;
        end
    end

    for i = 1:ntr
        clabels = trOutput{i};
        save([str 'slice' num2str(i)],'clabels','-v7.3');
    end
%     matlabpool close;
else
    fprintf('Outputs of stage %d level %d already generated...skip \n',stage,level);
end

Answer 1

[Fangjun Jiang]

Here is a simple way to duplicate your error. Hope you will get it. When you have a matrix of say 2x3 elements, you can't access the (2,4) element because it does not exist.

>> clear all
>> fv=ones(2,68160);
>> indSu=68161;
>> fv=fv(:,indSu)
Index in position 2 exceeds array bounds (must not exceed 68160).

[敏敏 杜]

Thank you, I tried it but it didn't work...