IMDBAddMIT67.m

function [ imdb, select ] = IMDBAddMIT67( imdb, path, task_num, varargin )
% IMDBADDMIT67   Add to imdb the MIT indoor scene dataset.
%   Input:
%     PATH struct generated by GETPATH() with path.path_OrigNetresponse field adjusted
%     TASK_NUM the path number (see CNN_CUSTOMTRAIN) that this dataset corresponds to
%   Options:
%     See code comments
% 
% Authors: Zhizhong Li
% 
% See the COPYING file.

opts.partial = 0; % for >0 partial, e.g. 0.3, only include that much portion of # samples.
opts.label = 'class'; % 'class' for class label, 'probdist' for a recorded response distribution (uses path.path_OrigNetresponse.MIT67)
opts.trainval = [1 2]; % 1 for train, 2 for val. By default include train+val.
opts.randstream = []; % use randstream if provided
opts = vl_argparse(opts, varargin);


if ~isfield(imdb, 'images')
    imdb.images.name = [];
    imdb.images.label = [];
    imdb.images.set = [];
    imdb.images.task = [];
end

fid = fopen(fullfile(path.path_MIT67root, 'classes.txt'), 'r');
classstrs = textscan(fid, '%s\n'); classstrs = classstrs{1};
fclose(fid);
classmap = containers.Map(classstrs, num2cell(1:numel(classstrs)));

sets = {'train', 'test'}; sets = sets(opts.trainval);

% train or test set
for f = sets
    f = char(f);
    set = 1; if strcmp(f, 'test'), set = 2; end

    % image names
    itemread = fileread(fullfile(path.path_MIT67meta, [f 'Images.txt']));
    readname.(f){1} = strsplit(itemread, {'\n'})';
    itemclss = strsplit(itemread, {'\n', '/'})'; itemclss = itemclss(1:2:end);
    itemclss = values(classmap, itemclss);

    % selecting partial
    if opts.partial
        if numel(opts.partial)==1
            partial = opts.partial;
        else
            partial = opts.partial(set);
        end
        if isempty(opts.randstream)
            select.(f) = randperm(numel(readname.(f){1}), ceil(numel(readname.(f){1}) * partial));
        else
            select.(f) = randperm(opts.randstream, numel(readname.(f){1}), ceil(numel(readname.(f){1}) * partial));
        end
    else
        select.(f) = 1:numel(readname.(f){1});
    end

    % names
    names = readname.(f){1}(select.(f));
    n_set = size(names,1);

    % labels...
    switch opts.label
        case 'probdist'
            % distribution: just load
            probdist = load(path.path_OrigNetresponse.MIT67.(f));
            classes = num2cell( probdist.lastfc_out, 1 )';
        case 'class'
            % induct from file folder name
            classes = itemclss;
        otherwise
            throw(MException('opts.label:notRecognized', 'probdist/class'));
    end

    classes = classes(select.(f),:);
    
    imdb.images.name = [ imdb.images.name; names ];
    imdb.images.label = [ imdb.images.label; classes ];

    imdb.images.set = [ imdb.images.set;
        ones(n_set,1) * set ];
    imdb.images.task = [ imdb.images.task;
        task_num * ones(n_set, 1) ];
end

if isfield(select, 'test'), select.val = select.test; select = rmfield(select, 'test'); end