metadata_editor.cpp 16.2 KB
Newer Older
1
2
3
4
5
6
// Copyright (C) 2011  Davis E. King (davis@dlib.net)
// License: Boost Software License   See LICENSE.txt for the full license.

#include "metadata_editor.h"
#include <dlib/array.h>
#include <dlib/queue.h>
7
#include <dlib/static_set.h>
8
9
10
11
#include <dlib/misc_api.h>
#include <dlib/image_io.h>
#include <dlib/array2d.h>
#include <dlib/pixel.h>
Davis King's avatar
Davis King committed
12
#include <sstream>
13
14
15
16

using namespace std;
using namespace dlib;

Davis King's avatar
Davis King committed
17
18
extern const char* VERSION;

19
20
21
22
23
24
25
26
// ----------------------------------------------------------------------------------------

metadata_editor::
metadata_editor(
    const std::string& filename_
) : 
    mbar(*this),
    lb_images(*this),
27
28
29
30
    image_pos(0),
    display(*this),
    overlay_label_name(*this),
    overlay_label(*this)
31
{
32
33
    file metadata_file(filename_);
    filename = metadata_file.full_name();
34
35
36
    // Make our current directory be the one that contains the metadata file.  We 
    // do this because that file might contain relative paths to the image files
    // we are supposed to be loading.
37
38
    set_current_dir(get_parent_directory(metadata_file).full_name());

39
40
41
42
43
44
45
46
47
48
49
50
51
    load_image_dataset_metadata(metadata, filename);

    dlib::array<std::string>::expand_1a files;
    files.resize(metadata.images.size());
    for (unsigned long i = 0; i < metadata.images.size(); ++i)
    {
        files[i] = metadata.images[i].filename;
    }
    lb_images.load(files);
    lb_images.enable_multiple_select();

    lb_images.set_click_handler(*this, &metadata_editor::on_lb_images_clicked);

52
53
    overlay_label_name.set_text("Next Label: ");
    overlay_label.set_width(200);
54

55
    display.set_overlay_rects_changed_handler(*this, &metadata_editor::on_overlay_rects_changed);
56
    display.set_overlay_rect_selected_handler(*this, &metadata_editor::on_overlay_rect_selected);
57
58
    overlay_label.set_text_modified_handler(*this, &metadata_editor::on_overlay_label_changed);

Davis King's avatar
Davis King committed
59
    mbar.set_number_of_menus(2);
60
    mbar.set_menu_name(0,"File",'F');
Davis King's avatar
Davis King committed
61
    mbar.set_menu_name(1,"Help",'H');
62
63
64
65
66
67
68


    mbar.menu(0).add_menu_item(menu_item_text("Save",*this,&metadata_editor::file_save,'S'));
    mbar.menu(0).add_menu_item(menu_item_text("Save As",*this,&metadata_editor::file_save_as,'A'));
    mbar.menu(0).add_menu_item(menu_item_separator());
    mbar.menu(0).add_menu_item(menu_item_text("Remove Selected Images",*this,&metadata_editor::remove_selected_images,'R'));
    mbar.menu(0).add_menu_item(menu_item_separator());
69
    mbar.menu(0).add_menu_item(menu_item_text("Exit",static_cast<base_window&>(*this),&drawable_window::close_window,'x'));
70

Davis King's avatar
Davis King committed
71
    mbar.menu(1).add_menu_item(menu_item_text("About",*this,&metadata_editor::display_about,'A'));
72

73
    // set the size of this window.
74
    on_window_resized();
Davis King's avatar
Davis King committed
75
    load_image_and_set_size(0);
76
77
78
    on_window_resized();
    if (image_pos < lb_images.size() )
        lb_images.select(image_pos);
79

80
81
82
83
84
85
86
87
    // make sure the window is centered on the screen.
    unsigned long width, height;
    get_size(width, height);
    unsigned long screen_width, screen_height;
    get_display_size(screen_width, screen_height);
    set_pos((screen_width-width)/2, (screen_height-height)/2);

    set_title("Image Labeler - " + metadata.name);
88
89
90
91
92
93
94
95
96
97
98
99
100
101
    show();
} 

// ----------------------------------------------------------------------------------------

metadata_editor::
~metadata_editor(
)
{
    close_window();
}

// ----------------------------------------------------------------------------------------

102
103
104
105
106
107
108
109
110
111
void metadata_editor::
add_labelable_part_name (
    const std::string& name
)
{
    display.add_labelable_part_name(name);
}

// ----------------------------------------------------------------------------------------

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
141
142
143
144
145
146
147
void metadata_editor::
file_save()
{
    save_metadata_to_file(filename);
}

// ----------------------------------------------------------------------------------------

void metadata_editor::
save_metadata_to_file (
    const std::string& file
)
{
    try
    {
        save_image_dataset_metadata(metadata, file);
    }
    catch (dlib::error& e)
    {
        message_box("Error saving file", e.what());
    }
}

// ----------------------------------------------------------------------------------------

void metadata_editor::
file_save_as()
{
    save_file_box(*this, &metadata_editor::save_metadata_to_file);
}

// ----------------------------------------------------------------------------------------

void metadata_editor::
remove_selected_images()
{
148
149
150
    dlib::queue<unsigned long>::kernel_1a list;
    lb_images.get_selected(list);
    list.reset();
151
    unsigned long min_idx = lb_images.size();
152
153
154
    while (list.move_next())
    {
        lb_images.unselect(list.element());
155
        min_idx = std::min(min_idx, list.element());
156
157
    }

158

159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
    // remove all the selected items from metadata.images
    dlib::static_set<unsigned long>::kernel_1a to_remove;
    to_remove.load(list);
    std::vector<dlib::image_dataset_metadata::image> images;
    for (unsigned long i = 0; i < metadata.images.size(); ++i)
    {
        if (to_remove.is_member(i) == false)
        {
            images.push_back(metadata.images[i]);
        }
    }
    images.swap(metadata.images);


    // reload metadata into lb_images
    dlib::array<std::string>::expand_1a files;
    files.resize(metadata.images.size());
    for (unsigned long i = 0; i < metadata.images.size(); ++i)
    {
        files[i] = metadata.images[i].filename;
    }
    lb_images.load(files);


183
184
185
    if (min_idx != 0)
        min_idx--;
    select_image(min_idx);
186
187
188
189
190
191
192
193
194
195
196
197
198
199
}

// ----------------------------------------------------------------------------------------

void metadata_editor::
on_window_resized(
)
{
    drawable_window::on_window_resized();

    unsigned long width, height;
    get_size(width, height);

    lb_images.set_pos(0,mbar.bottom()+1);
200
    lb_images.set_size(180, height - mbar.height());
201
202
203
204
205
206

    overlay_label_name.set_pos(lb_images.right()+10, mbar.bottom() + (overlay_label.height()-overlay_label_name.height())/2+1);
    overlay_label.set_pos(overlay_label_name.right(), mbar.bottom()+1);
    display.set_pos(lb_images.right(), overlay_label.bottom()+3);

    display.set_size(width - display.left(), height - display.top());
207
208
209
210
}

// ----------------------------------------------------------------------------------------

211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
void propagate_labels(
    const std::string& label,
    dlib::image_dataset_metadata::dataset& data,
    unsigned long prev,
    unsigned long next
)
{
    if (prev == next || next >= data.images.size())
        return;


    for (unsigned long i = 0; i < data.images[prev].boxes.size(); ++i)
    {
        if (data.images[prev].boxes[i].label != label)
            continue;

        // figure out which box in the next image matches the current one the best
        const rectangle cur = data.images[prev].boxes[i].rect;
        double best_overlap = 0;
        unsigned long best_idx = 0;
        for (unsigned long j = 0; j < data.images[next].boxes.size(); ++j)
        {
            const rectangle next_box = data.images[next].boxes[j].rect;
            const double overlap = cur.intersect(next_box).area()/(double)(cur+next_box).area();
            if (overlap > best_overlap)
            {
                best_overlap = overlap;
                best_idx = j;
            }
        }

        // If we found a matching rectangle in the next image and the best match doesn't
        // already have a label.
        if (best_overlap > 0.5 && data.images[next].boxes[best_idx].label == "")
        {
            data.images[next].boxes[best_idx].label = label;
        }
    }

}

252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
// ----------------------------------------------------------------------------------------

bool has_label_or_all_boxes_labeled (
    const std::string& label,
    const dlib::image_dataset_metadata::image& img 
)
{
    if (label.size() == 0)
        return true;

    bool all_boxes_labeled = true;
    for (unsigned long i = 0; i < img.boxes.size(); ++i)
    {
        if (img.boxes[i].label == label)
            return true;
        if (img.boxes[i].label.size() == 0)
            all_boxes_labeled = false;
    }

    return all_boxes_labeled;
}

// ----------------------------------------------------------------------------------------

276
277
278
279
280
281
282
283
284
285
void metadata_editor::
on_keydown (
    unsigned long key,
    bool is_printable,
    unsigned long state
)
{
    drawable_window::on_keydown(key, is_printable, state);

    if (is_printable)
286
287
288
289
290
291
292
    {
        if (key == '\t')
        {
            overlay_label.give_input_focus();
            overlay_label.select_all_text();
        }

293
        return;
294
    }
295
296
297

    if (key == base_window::KEY_UP)
    {
298
        if (state&base_window::KBD_MOD_CONTROL)
299
300
301
302
303
304
305
306
307
308
        {
            // If the label we are supposed to propagate doesn't exist in the current image
            // then don't advance.
            if (!has_label_or_all_boxes_labeled(display.get_default_overlay_rect_label(),metadata.images[image_pos]))
                return;

            // if the next image is going to be empty then fast forward to the next one
            while (image_pos > 1 && metadata.images[image_pos-1].boxes.size() == 0)
                --image_pos;

309
            propagate_labels(display.get_default_overlay_rect_label(), metadata, image_pos, image_pos-1);
310
        }
311
312
313
314
        select_image(image_pos-1);
    }
    else if (key == base_window::KEY_DOWN)
    {
315
        if (state&base_window::KBD_MOD_CONTROL)
316
317
318
319
320
321
322
323
324
325
        {
            // If the label we are supposed to propagate doesn't exist in the current image
            // then don't advance.
            if (!has_label_or_all_boxes_labeled(display.get_default_overlay_rect_label(),metadata.images[image_pos]))
                return;

            // if the next image is going to be empty then fast forward to the next one
            while (image_pos+1 < metadata.images.size() && metadata.images[image_pos+1].boxes.size() == 0)
                ++image_pos;

326
            propagate_labels(display.get_default_overlay_rect_label(), metadata, image_pos, image_pos+1);
327
        }
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
        select_image(image_pos+1);
    }
}

// ----------------------------------------------------------------------------------------

void metadata_editor::
select_image(
    unsigned long idx
)
{
    if (idx < lb_images.size())
    {
        // unselect all currently selected images
        dlib::queue<unsigned long>::kernel_1a list;
        lb_images.get_selected(list);
        list.reset();
        while (list.move_next())
        {
            lb_images.unselect(list.element());
        }


        lb_images.select(idx);
352
        load_image(idx);
353
    }
Davis King's avatar
Davis King committed
354
    else if (lb_images.size() == 0)
Davis King's avatar
Davis King committed
355
356
357
358
359
    {
        display.clear_overlay();
        array2d<unsigned char> empty_img;
        display.set_image(empty_img);
    }
360
361
362
363
364
365
366
367
368
}

// ----------------------------------------------------------------------------------------

void metadata_editor::
on_lb_images_clicked(
    unsigned long idx
) 
{ 
369
    load_image(idx);
370
371
372
373
}

// ----------------------------------------------------------------------------------------

Davis King's avatar
Davis King committed
374
375
376
377
378
379
380
381
382
std::vector<dlib::image_display::overlay_rect> get_overlays (
    const dlib::image_dataset_metadata::image& data
)
{
    std::vector<dlib::image_display::overlay_rect> temp(data.boxes.size());
    for (unsigned long i = 0; i < temp.size(); ++i)
    {
        temp[i].rect = data.boxes[i].rect;
        temp[i].label = data.boxes[i].label;
383
        temp[i].parts = data.boxes[i].parts;
384
        temp[i].crossed_out = data.boxes[i].ignore;
Davis King's avatar
Davis King committed
385
386
387
388
389
390
391
        assign_pixel(temp[i].color, rgb_pixel(255,0,0));
    }
    return temp;
}

// ----------------------------------------------------------------------------------------

392
393
394
395
396
397
398
399
void metadata_editor::
load_image(
    unsigned long idx
)
{
    if (idx >= metadata.images.size())
        return;

Davis King's avatar
Davis King committed
400
401
    image_pos = idx; 

402
403
404
405
406
407
408
409
410
411
412
413
    array2d<rgb_pixel> img;
    display.clear_overlay();
    try
    {
        dlib::load_image(img, metadata.images[idx].filename);

    }
    catch (exception& e)
    {
        message_box("Error loading image", e.what());
    }

414
    display.set_image(img);
Davis King's avatar
Davis King committed
415
    display.add_overlay(get_overlays(metadata.images[idx]));
416
417
418
419
420
421
422
423
424
425
426
427
}

// ----------------------------------------------------------------------------------------

void metadata_editor::
load_image_and_set_size(
    unsigned long idx
)
{
    if (idx >= metadata.images.size())
        return;

Davis King's avatar
Davis King committed
428
429
    image_pos = idx; 

430
431
432
433
434
435
436
437
438
439
440
441
    array2d<rgb_pixel> img;
    display.clear_overlay();
    try
    {
        dlib::load_image(img, metadata.images[idx].filename);

    }
    catch (exception& e)
    {
        message_box("Error loading image", e.what());
    }

442

443
444
    unsigned long screen_width, screen_height;
    get_display_size(screen_width, screen_height);
445

446

447
448
449
450
451
452
453
454
455
    unsigned long needed_width = display.left() + img.nc() + 4;
    unsigned long needed_height = display.top() + img.nr() + 4;
	if (needed_width < 300) needed_width = 300;
	if (needed_height < 300) needed_height = 300;

    if (needed_width+50 < screen_width &&
        needed_height+50 < screen_height)
    {
        set_size(needed_width, needed_height);
456
457
    }

458

459
    display.set_image(img);
Davis King's avatar
Davis King committed
460
    display.add_overlay(get_overlays(metadata.images[idx]));
461
462
463
464
}

// ----------------------------------------------------------------------------------------

465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
void metadata_editor::
on_overlay_rects_changed(
)
{
    using namespace dlib::image_dataset_metadata;
    if (image_pos < metadata.images.size())
    {
        const std::vector<image_display::overlay_rect>& rects = display.get_overlay_rects();

        std::vector<box>& boxes = metadata.images[image_pos].boxes;

        boxes.clear();
        for (unsigned long i = 0; i < rects.size(); ++i)
        {
            box temp;
            temp.label = rects[i].label;
            temp.rect = rects[i].rect;
482
            temp.parts = rects[i].parts;
483
            temp.ignore = rects[i].crossed_out;
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
            boxes.push_back(temp);
        }
    }
}

// ----------------------------------------------------------------------------------------

void metadata_editor::
on_overlay_label_changed(
)
{
    display.set_default_overlay_rect_label(trim(overlay_label.text()));
}

// ----------------------------------------------------------------------------------------
499

500
501
502
503
504
505
506
507
508
509
510
void metadata_editor::
on_overlay_rect_selected(
    const image_display::overlay_rect& orect
)
{
    overlay_label.set_text(orect.label);
    display.set_default_overlay_rect_label(orect.label);
}

// ----------------------------------------------------------------------------------------

Davis King's avatar
Davis King committed
511
512
513
514
515
516
517
518
void metadata_editor::
display_about(
)
{
    std::ostringstream sout;
    sout << wrap_string("Image Labeler v" + string(VERSION) + "." ,0,0) << endl << endl;
    sout << wrap_string("This program is a tool for labeling images with rectangles. " ,0,0) << endl << endl;

519
520
521
522
523
524
525
    sout << wrap_string("You can add a new rectangle by holding the shift key, left clicking "
                        "the mouse, and dragging it.  New rectangles are given the label from the \"Next Label\" "
                        "field at the top of the application.  You can quickly edit the contents of the Next Label field "
                        "by hitting the tab key. Double clicking "
                        "a rectangle selects it and the delete key removes it.  You can also mark "
                        "a rectangle as ignored by hitting the i key when it is selected.  Ignored "
                        "rectangles are visually displayed with an X through them."
Davis King's avatar
Davis King committed
526
527
                        ,0,0) << endl << endl;

528
529
530
531
532
533
    sout << wrap_string("It is also possible to label object parts by selecting a rectangle and "
                        "then right clicking.  A popup menu will appear and you can select a part label. "
                        "Note that you must define the allowable part labels by giving --parts on the "
                        "command line.  An example would be '--parts \"leye reye nose mouth\"'."
                        ,0,0) << endl << endl;

534
    sout << wrap_string("Additionally, you can hold ctrl and then scroll the mouse wheel to zoom.  A normal left click "
535
536
537
538
                        "and drag allows you to navigate around the image.  Holding ctrl and "
                        "left clicking a rectangle will give it the label from the Next Label field. "
                        "Finally, holding ctrl and pressing the up or down keyboard keys will propagate "
                        "rectangle labels from one image to the next.",0,0) << endl;
Davis King's avatar
Davis King committed
539
540
541
542
543
544

    message_box("About Image Labeler",sout.str());
}

// ----------------------------------------------------------------------------------------