dan@9: function [trainlists, testlists, testgtlists] = sceneClassificationMetrics_createFolds(wavsourcedir, targetdir)
dan@9: 
dan@9: % [trainlists, testlists, testgtlists] = sceneClassificationMetrics_createFolds('~/data/aasp_chall_FROMRDR2/scenes_stereo/scenes_stereo', 'tmp')
dan@9: % [trainlists, testlists, testgtlists] = sceneClassificationMetrics_createFolds('scenes_stereo', 'tmp');
dan@9: 
dan@9: numfolds=5;
dan@9: 
dan@9: wavpaths = dir(fullfile(wavsourcedir, '*.wav'));
dan@9: 
dan@9: % Slightly strange way to index class labels; it's because containers.Map not present in Octave
dan@9: classIDs.bus           =  1;
dan@9: classIDs.busystreet    =  2;
dan@9: classIDs.office        =  3;
dan@9: classIDs.openairmarket =  4;
dan@9: classIDs.park          =  5;
dan@9: classIDs.quietstreet   =  6;
dan@9: classIDs.restaurant    =  7;
dan@9: classIDs.supermarket   =  8;
dan@9: classIDs.tube          =  9;
dan@9: classIDs.tubestation   = 10;
dan@9: classcollections = cell(10,1);
dan@9: 
dan@9: % For every file, we create an entry in classcollections[id][], also allocating its fold number
dan@9: for index = 1:length(wavpaths)
dan@9:     wavpathobj = wavpaths(index);
dan@9:     wavitem.origpath  = fullfile(wavsourcedir, wavpathobj.name);
dan@9:     wavitem.classstr  = wavpathobj.name(:,1:length(wavpathobj.name)-6);  % rm 6 chars eg '09.wav'
dan@9:     wavitem.classID   = getfield(classIDs, wavitem.classstr);
dan@9:     numsofar = length(classcollections{wavitem.classID});
dan@9:     wavitem.whichfold = mod(numsofar, numfolds) + 1;
dan@9:     classcollections{wavitem.classID}{numsofar+1} = wavitem;
dan@9: end
dan@9: 
dan@9: 
dan@9: % for each fold, we'll create a folder with a scrambley name, and we'll initialise output files
dan@9: for whichfold=1:numfolds
dan@9:     foldfolders{whichfold} = fullfile(targetdir, sprintf('%d_%s', whichfold, RandomString1(6)));
dan@9:     mkdir(foldfolders{whichfold});
dan@9:     trainlists{whichfold}  = fullfile(foldfolders{whichfold}, sprintf('fold%d_train.txt', whichfold));
dan@9:     testlists{whichfold}   = fullfile(foldfolders{whichfold}, sprintf('fold%d_test.txt', whichfold));
dan@9:     testgtlists{whichfold} = fullfile(targetdir,              sprintf('fold%d_testgt_%s.txt', whichfold, RandomString1(6)));
dan@9:     f_trainlists{whichfold}  = fopen(trainlists{ whichfold}, 'w');
dan@9:     f_testlists{whichfold}   = fopen(testlists{  whichfold}, 'w');
dan@9:     f_testgtlists{whichfold} = fopen(testgtlists{whichfold}, 'w');
dan@9: end
dan@9: 
dan@9: for whichclass = 1:length(classcollections)
dan@9:     thisclass = classcollections{whichclass};
dan@9:     
dan@9:     for whichitem = 1:length(thisclass)
dan@9:         wavitem = thisclass{whichitem};
dan@9:         for whichfold=1:numfolds
dan@9:             newfilename = fullfile(foldfolders{whichfold}, sprintf('%d_%s.wav', whichitem, RandomString1(6)));
dan@9:             copyfile(wavitem.origpath, newfilename); % BE CAREFUL WITH THIS!!!
dan@9:             if whichfold==wavitem.whichfold
dan@9:                 % this fold is TESTING for this item
dan@9:                 fprintf(f_testgtlists{whichfold}, '%s\t%s\n', newfilename, wavitem.classstr);
dan@9:                 fprintf(f_testlists{  whichfold}, '%s\n', newfilename);
dan@9:             else
dan@9:                 % this fold is TRAINING for this item
dan@9:                 fprintf(f_trainlists{whichfold}, '%s\t%s\n', newfilename, wavitem.classstr);
dan@9:             end
dan@9:         end
dan@9:     end
dan@9: end
dan@9: 
dan@9: % close the files
dan@9: for whichfold=1:numfolds
dan@9:     fclose(f_trainlists{ whichfold});
dan@9:     fclose(f_testlists{  whichfold});
dan@9:     fclose(f_testgtlists{whichfold});
dan@9: end
dan@9: 
dan@9: