File size: 5,924 Bytes
d4035c1 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 |
function LMcookdatabase(database, HOMEIMAGES, HOMEANNOTATIONS, NEWHOMEIMAGES, NEWHOMEANNOTATIONS, varargin)
% You can "cook" your own database by reformating the images and the
% annotation files so that the images in the database have some
% characteristics of interest. For instance, you might want a database in
% which all the images are smaller that 256 x 256. Or you might want a
% database of images that have cars, so that each car has a size of 32x32 pixels.
%
% This function captures some of the most common charcteristics for
% standarizing the databases.
%
% List of parameters
% -----------------------
% 'imagesize' [nrows ncols]
%
% 'objectname' defines which will be the object of interest
%
% 'objectsize' A number between 1 and inf; It is the size of the
% object of interest in the final image.
%
% 'objectlocation' Defines the location of the object in the new image.
% One of these strings: 'original', 'random', 'centered'
%
% 'multiplecrops' One of these strings: 'yes', 'no'.
% If there are multiple instances of the object of
% interest, it will create one image for each instance.
%
% Example:
% The next code generates a new database with street scenes. In the final
% database, there are only images with one car in each image. The images
% will have a maximal size of 256x256 pixels, and the car will have a fixed
% size of 64x64 (tight to the longest dimension).
%
% %Define the root folder for the images
% HOMEIMAGES = 'C:\atb\Databases\CSAILobjectsAndScenes\Images'; % you can set here your default folder
% HOMEANNOTATIONS = 'C:\atb\DATABASES\LabelMe\Annotations'; % you can set here your default folder
%
% %Define the root folder for the images
% NEWHOMEIMAGES = 'C:\atb\Projects\objectsAndScenes\generativeModels\iccv2005\database\streets\Images'; % you can set here your default folder
% NEWHOMEANNOTATIONS = 'C:\atb\Projects\objectsAndScenes\generativeModels\iccv2005\database\streets\Annotations'; % you can set here your default folder
%
% %Create database
% database = LMdatabase(HOMEANNOTATIONS);
%
% %Locate street scenes with only one car.
% [D,j] = LMquery(database, 'object.name', 'car+side');
% counts = LMcountobject(database(j), 'car');
% j = j(find(counts==1)); length(j)
% D = LMquery(database(j), 'object.name', 'car+side,building,road');
% LMdbshowscenes(D, HOMEIMAGES);
%
% %Cook database to fit our requirements: objects of fixed size
% LMcookdatabase(D, HOMEIMAGES, HOMEANNOTATIONS, NEWHOMEIMAGES, NEWHOMEANNOTATIONS, ...
% 'objectname', 'car', 'objectsize', [64 64], 'objectlocation', 'original', 'maximagesize', [256 256])
variables = {'maximagesize', 'objectname', 'objectsize', 'objectlocation', 'multiplecrops', 'impad', 'minobjectsize', 'polygoninstance'};
defaults = {[], '', [], 'original', 'no', [], [], []};
[maximagesize, objectname, objectsize, objectlocation, multiplecrops, impad, minobjectsize, polygoninstance] = ...
parseparameters(varargin, variables, defaults);
% Create structure of files:
Nfiles = length(database);
for i = 1:Nfiles
folders{i} = database(i).annotation.folder;
end
folders = unique(folders);
Nfolders = length(folders);
if length(dir(NEWHOMEIMAGES))>0; error(sprintf('Error: The folder %s already exists.', NEWHOMEIMAGES)); end
if length(dir(NEWHOMEANNOTATIONS))>0; error(sprintf('Error: The folder %s already exists.', NEWHOMEANNOTATIONS)); end
for i = 1:Nfolders
mkdir(NEWHOMEIMAGES, folders{i});
mkdir(NEWHOMEANNOTATIONS, folders{i});
end
% Save readme file in the root of the NEWHOMEANNOTATIONS and NEWHOMEIMAGES
readme{1} = 'This database is a subset of the LabelMe dataset.';
readme{2} = 'http://people.csail.mit.edu/brussell/research/LabelMe/intro.html';
readme{3} = '';
readme{4} = 'This dataset has been created with the parameters:';
for n=1:length(varargin); readme{4+n} = varargin{n}; end
fid = fopen(fullfile(NEWHOMEIMAGES, 'readme.txt'), 'w');
for i = 1:length(readme)
fprintf(fid, '%s \n', readme{i});
end
fclose(fid)
% Create new database
for n = 1:Nfiles
clear newannotation newimg err
img = LMimread(database, n, HOMEIMAGES); % Load image
if strcmp(multiplecrops, 'yes')
% we will do the image manipulation by centering on each instance.
Nimages = length(database(n).annotation.object);
for m = 1:Nimages
[newannotation{m}, newimg{m}, crop, scaling, err{m}, msg] = LMcookimage(database(n).annotation, img, varargin{:}, 'polygoninstance', m);
end
else
Nimages = 1;
[newannotation{1}, newimg{1}, crop, scaling, err{1}, msg] = LMcookimage(database(n).annotation, img, varargin{:});
end
% The images are saved only if there was no upsampling (no blur)
for m = 1:Nimages
if err{m} == 0
% Save image and annotation
if Nimages==1
filename = database(n).annotation.filename;
else
filename = sprintf('%s_crop%d.jpg', strrep(database(n).annotation.filename, '.jpg', ''), m);
end
% write new image
imwrite(newimg{m}, fullfile(NEWHOMEIMAGES, database(n).annotation.folder, filename), 'jpg', 'quality', 100);
% write new annotation file
v.annotation = newannotation{m};
v.annotation.filename = filename;
writeXML(fullfile(NEWHOMEANNOTATIONS, database(n).annotation.folder, strrep(filename,'.jpg','.xml')), v);
figure(1); clf
subplottight(1,2,1)
LMplot(database(n).annotation, img)
subplottight(1,2,2)
LMplot(newannotation{m}, newimg{m})
title(sprintf('%d/%d', n, Nfiles))
xlabel(strrep(fullfile(database(n).annotation.folder, filename), '\', '/'))
drawnow
end
end
end
|