File size: 25,283 Bytes
7471ead
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
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
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
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
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
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
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
# python create_dataset.py
# cd .. && huggingface-cli upload-large-folder anokimchen/geometric_shapes geometric_shapes  --repo-type dataset && cd geometric_shapes


# Python version: 3.10.10 (main, Mar 21 2023, 13:41:39) [Clang 14.0.6 ]
# os: Built-in (part of Python 3.10.10)
# random: Built-in (part of Python 3.10.10)
# csv: 1.0
# numpy: 1.26.4
# matplotlib: 3.8.2
# tqdm: 4.67.1
# time: Built-in (part of Python 3.10.10)
# json: 2.0.9
# multiprocessing: Built-in (part of Python 3.10.10)

import os
import random
import csv
import numpy as np
import matplotlib
matplotlib.use('Agg')  # Set non-interactive backend
import matplotlib.pyplot as plt
from matplotlib.path import Path
import matplotlib.patches as patches
from tqdm import tqdm
import time, json
import multiprocessing as mp
from functools import partial

# List of shapes
shapes = [
    "circle", "square", "triangle", "rectangle", "pentagon", "hexagon", 
    "heptagon", "octagon", "nonagon", "decagon", "hendecagon", "dodecagon", 
    "tridecagon", "tetradecagon", "pentadecagon", "hexadecagon", "heptadecagon", 
    "octadecagon", "enneadecagon", "icosagon", "ellipse", "parallelogram", 
    "trapezoid", "rhombus", "star", "crescent", "heart", "cross", "arrow", 
    "diamond", "kite", "oval", "semicircle", "sector", "torus", "annulus", 
    "deltoid", "astroid", "superellipse", "hypotrochoid", "epitrochoid", 
    "lemniscate", "quadrifolium", "trefoil", "clover", "bean", "peanut", 
    "lune", "vesica piscis", "spherical cap", "spherical wedge", 
    "spherical lune", "hypocycloid", "epicycloid"
]

# Colors with their names
colors = {
    'red': '#FF0000',
    'green': '#00FF00',
    'blue': '#0000FF',
    'yellow': '#FFFF00',
    'cyan': '#00FFFF',
    'magenta': '#FF00FF',
    'purple': '#800080',
    'orange': '#FFA500',
    'pink': '#FFC0CB',
    'brown': '#A52A2A',
    'black': '#000000',
    'white': '#FFFFFF',
    'gray': '#808080',
    'lime': '#32CD32',
    'navy': '#000080',
    'teal': '#008080',
    'olive': '#808000',
    'maroon': '#800000',
    'coral': '#FF7F50',
    'gold': '#FFD700',
    'silver': '#C0C0C0',
    'indigo': '#4B0082',
    'turquoise': '#40E0D0',
    'violet': '#EE82EE',
    'khaki': '#F0E68C',
    'salmon': '#FA8072',
    'crimson': '#DC143C',
    'lavender': '#E6E6FA',
    'plum': '#DDA0DD',
    'orchid': '#DA70D6',
    'chocolate': '#D2691E'
}

# Size descriptions
size_descriptions = {
    (0, 15): ["tiny", "minuscule", "microscopic", "very small", "itty-bitty"],
    (15, 30): ["small", "little", "petite", "diminutive", "compact"],
    (30, 50): ["medium", "moderate", "average", "intermediate", "mid-sized"],
    (50, 75): ["large", "big", "substantial", "sizable", "considerable"],
    (75, 90): ["very large", "huge", "enormous", "immense", "massive"],
    (90, 100): ["gigantic", "colossal", "gargantuan", "titanic", "mammoth"]
}

# Define polygon sides mapping at module level
polygon_sides = {
    "pentagon": 5, "hexagon": 6, "heptagon": 7, "octagon": 8, "nonagon": 9, "decagon": 10,
    "hendecagon": 11, "dodecagon": 12, "tridecagon": 13, "tetradecagon": 14, 
    "pentadecagon": 15, "hexadecagon": 16, "heptadecagon": 17, "octadecagon": 18, 
    "enneadecagon": 19, "icosagon": 20
}

def get_size_description(size_percent):
    for (min_val, max_val), descriptions in size_descriptions.items():
        if min_val <= size_percent < max_val:
            return random.choice(descriptions)
    return random.choice(size_descriptions[(90, 100)])  # Default to largest if out of range

def regular_polygon(sides, radius=0.4, rotation=0, center=(0.5, 0.5)):
    """Generate coordinates for a regular polygon."""
    angles = np.linspace(0 + rotation, 2 * np.pi + rotation, sides + 1)[:-1]
    x = center[0] + radius * np.cos(angles)
    y = center[1] + radius * np.sin(angles)
    return list(zip(x, y))

def star_coords(points=5, inner_radius=0.2, outer_radius=0.4, center=(0.5, 0.5)):
    """Generate coordinates for a star."""
    all_angles = np.linspace(0, 2 * np.pi, 2 * points, endpoint=False)
    radii = [outer_radius, inner_radius] * points
    x = center[0] + np.array([r * np.cos(a) for r, a in zip(radii, all_angles)])
    y = center[1] + np.array([r * np.sin(a) for r, a in zip(radii, all_angles)])
    return list(zip(x, y))

def draw_shape(shape, ax, size_factor=1.0):
    """Draw a shape with the given size factor (0.0 to 1.0)."""
    center = (0.5, 0.5)
    color_name = random.choice(list(colors.keys()))
    color_hex = colors[color_name]
    
    # Base size for scaling (as a fraction of available space)
    base_size = 0.4 * size_factor
    
    if shape == "circle":
        circle = plt.Circle(center, base_size, color=color_hex)
        ax.add_patch(circle)
    
    elif shape == "square":
        offset = 0.5 - base_size
        square = plt.Rectangle((offset, offset), 2*base_size, 2*base_size, color=color_hex)
        ax.add_patch(square)
    
    elif shape == "triangle":
        height = 2 * base_size
        base_width = 2 * base_size
        triangle = plt.Polygon([
            (0.5, 0.5 + height/2),
            (0.5 - base_width/2, 0.5 - height/2),
            (0.5 + base_width/2, 0.5 - height/2)
        ], color=color_hex)
        ax.add_patch(triangle)
    
    elif shape == "rectangle":
        width = 2 * base_size
        height = 1.5 * base_size
        offset_x = 0.5 - width/2
        offset_y = 0.5 - height/2
        rectangle = plt.Rectangle((offset_x, offset_y), width, height, color=color_hex)
        ax.add_patch(rectangle)
    
    elif shape in polygon_sides:
        sides = polygon_sides[shape]
        polygon = plt.Polygon(regular_polygon(sides, radius=base_size), color=color_hex)
        ax.add_patch(polygon)
    
    elif shape == "ellipse" or shape == "oval":
        width = 2 * base_size
        height = 1.3 * base_size
        ellipse = patches.Ellipse(center, width, height, color=color_hex)
        ax.add_patch(ellipse)
    
    elif shape == "parallelogram":
        skew = base_size * 0.3
        parallelogram = plt.Polygon([
            (0.5-base_size, 0.5-base_size/2),
            (0.5+base_size-skew, 0.5-base_size/2),
            (0.5+base_size, 0.5+base_size/2),
            (0.5-base_size+skew, 0.5+base_size/2)
        ], color=color_hex)
        ax.add_patch(parallelogram)
    
    elif shape == "trapezoid":
        top_width = base_size * 1.2
        bottom_width = base_size * 2
        height = base_size * 1.5
        trapezoid = plt.Polygon([
            (0.5-bottom_width/2, 0.5-height/2),
            (0.5+bottom_width/2, 0.5-height/2),
            (0.5+top_width/2, 0.5+height/2),
            (0.5-top_width/2, 0.5+height/2)
        ], color=color_hex)
        ax.add_patch(trapezoid)
    
    elif shape == "rhombus" or shape == "diamond":
        width = base_size * 1.8
        height = base_size * 1.8
        rhombus = plt.Polygon([
            (0.5, 0.5-height/2),
            (0.5+width/2, 0.5),
            (0.5, 0.5+height/2),
            (0.5-width/2, 0.5)
        ], color=color_hex)
        ax.add_patch(rhombus)
    
    elif shape == "star":
        star = plt.Polygon(star_coords(
            points=5,
            inner_radius=base_size * 0.5,
            outer_radius=base_size * 1.3
        ), color=color_hex)
        ax.add_patch(star)
    
    elif shape == "crescent":
        # Create a crescent by taking the difference of two circles
        circle1 = plt.Circle(center, base_size, color=color_hex)
        offset = base_size * 0.3
        circle2 = plt.Circle((0.5 + offset, 0.5), base_size * 0.9, color='white')
        ax.add_patch(circle1)
        ax.add_patch(circle2)
    
    elif shape == "heart":
        t = np.linspace(0, 2*np.pi, 100)
        x = 0.5 + base_size * (16 * np.sin(t)**3) / 16
        y = 0.5 + base_size * (13 * np.cos(t) - 5 * np.cos(2*t) - 2 * np.cos(3*t) - np.cos(4*t)) / 16
        heart = plt.Polygon(list(zip(x, y)), color=color_hex)
        ax.add_patch(heart)
    
    elif shape == "cross":
        width = base_size * 0.5
        length = base_size * 1.5
        cross = plt.Polygon([
            (0.5-width/2, 0.5-length/2), (0.5+width/2, 0.5-length/2),
            (0.5+width/2, 0.5-width/2), (0.5+length/2, 0.5-width/2),
            (0.5+length/2, 0.5+width/2), (0.5+width/2, 0.5+width/2),
            (0.5+width/2, 0.5+length/2), (0.5-width/2, 0.5+length/2),
            (0.5-width/2, 0.5+width/2), (0.5-length/2, 0.5+width/2),
            (0.5-length/2, 0.5-width/2), (0.5-width/2, 0.5-width/2)
        ], color=color_hex)
        ax.add_patch(cross)
    
    elif shape == "arrow":
        arrow_length = base_size * 2
        arrow_width = base_size * 0.8
        arrow_head = base_size * 1.2
        arrow = plt.Polygon([
            (0.5-arrow_length/2, 0.5),
            (0.5+arrow_length/2-arrow_head, 0.5-arrow_width/2),
            (0.5+arrow_length/2-arrow_head, 0.5-arrow_width/4),
            (0.5+arrow_length/2, 0.5),
            (0.5+arrow_length/2-arrow_head, 0.5+arrow_width/4),
            (0.5+arrow_length/2-arrow_head, 0.5+arrow_width/2)
        ], color=color_hex)
        ax.add_patch(arrow)
    
    elif shape == "kite":
        width = base_size * 1.5
        height = base_size * 2.2
        kite = plt.Polygon([
            (0.5, 0.5+height/2),
            (0.5+width/2, 0.5),
            (0.5, 0.5-height/2),
            (0.5-width/2, 0.5)
        ], color=color_hex)
        ax.add_patch(kite)
    
    elif shape == "semicircle":
        theta = np.linspace(0, np.pi, 100)
        x = center[0] + base_size * np.cos(theta)
        y = center[1] + base_size * np.sin(theta)
        # Add a straight line at the bottom
        x = np.append(x, [center[0] - base_size, center[0]])
        y = np.append(y, [center[1], center[1]])
        semicircle = plt.Polygon(list(zip(x, y)), color=color_hex)
        ax.add_patch(semicircle)
    
    elif shape == "sector":
        theta = np.linspace(0, np.pi/2, 100)
        x = center[0] + base_size * np.cos(theta)
        y = center[1] + base_size * np.sin(theta)
        # Add lines to center
        x = np.append(x, [center[0]])
        y = np.append(y, [center[1]])
        sector = plt.Polygon(list(zip(x, y)), color=color_hex)
        ax.add_patch(sector)
    
    elif shape == "torus" or shape == "annulus":
        # Create an annulus/torus by drawing two concentric circles
        outer = plt.Circle(center, base_size, fill=True, color=color_hex)
        inner = plt.Circle(center, base_size * 0.5, fill=True, color='white')
        ax.add_patch(outer)
        ax.add_patch(inner)
    
    elif shape == "deltoid":
        t = np.linspace(0, 2*np.pi, 100)
        a, b = base_size, base_size * 0.3
        x = center[0] + a * np.cos(t) + b * np.cos(3*t)
        y = center[1] + a * np.sin(t) - b * np.sin(3*t)
        deltoid = plt.Polygon(list(zip(x, y)), color=color_hex)
        ax.add_patch(deltoid)
    
    elif shape == "astroid":
        t = np.linspace(0, 2*np.pi, 100)
        a = base_size * 1.3
        x = center[0] + a * np.cos(t)**3
        y = center[1] + a * np.sin(t)**3
        astroid = plt.Polygon(list(zip(x, y)), color=color_hex)
        ax.add_patch(astroid)
    
    elif shape == "lemniscate":
        t = np.linspace(-np.pi/4, np.pi/4, 100)
        a = base_size * 0.7
        x = center[0] + a * np.sqrt(2) * np.cos(t) / (np.sin(t)**2 + 1)
        y = center[1] + a * np.sqrt(2) * np.cos(t) * np.sin(t) / (np.sin(t)**2 + 1)
        
        # Add the bottom half
        t = np.linspace(3*np.pi/4, 5*np.pi/4, 100)
        x2 = center[0] + a * np.sqrt(2) * np.cos(t) / (np.sin(t)**2 + 1)
        y2 = center[1] + a * np.sqrt(2) * np.cos(t) * np.sin(t) / (np.sin(t)**2 + 1)
        
        x = np.append(x, x2)
        y = np.append(y, y2)
        
        plt.plot(x, y, color=color_hex, linewidth=4*size_factor)
    
    elif shape in ["quadrifolium", "trefoil", "clover"]:
        if shape == "quadrifolium":
            n = 4  # Four petals
        elif shape == "trefoil":
            n = 3  # Three petals
        else:  # clover
            n = 4  # Four petals, same as quadrifolium
            
        t = np.linspace(0, 2*np.pi, 1000)
        a = base_size
        x = center[0] + a * np.cos(n*t) * np.cos(t)
        y = center[1] + a * np.cos(n*t) * np.sin(t)
        plt.plot(x, y, color=color_hex, linewidth=4*size_factor)
    
    elif shape in ["bean", "peanut"]:
        t = np.linspace(0, 2*np.pi, 100)
        a, b = base_size * 1.5, base_size * 0.7
        x = center[0] + a * np.sin(t)
        y = center[1] + b * np.sin(t) * np.cos(t)
        bean = plt.Polygon(list(zip(x, y)), color=color_hex)
        ax.add_patch(bean)
    
    elif shape == "lune" or shape == "vesica piscis":
        # Create intersecting circles
        circle1 = plt.Circle((0.5 - base_size*0.3, 0.5), base_size, color=color_hex, alpha=0.7)
        circle2 = plt.Circle((0.5 + base_size*0.3, 0.5), base_size, color=color_hex, alpha=0.7)
        ax.add_patch(circle1)
        ax.add_patch(circle2)
    
    elif shape in ["hypocycloid", "epicycloid", "hypotrochoid", "epitrochoid"]:
        R = base_size * 1.2  # Radius of fixed circle
        r = base_size * 0.4  # Radius of rolling circle
        d = base_size * 0.2  # Distance from center of rolling circle
        
        t = np.linspace(0, 2*np.pi, 1000)
        
        if shape == "hypocycloid":
            x = center[0] + (R-r) * np.cos(t) + r * np.cos((R-r)*t/r)
            y = center[1] + (R-r) * np.sin(t) - r * np.sin((R-r)*t/r)
        elif shape == "epicycloid":
            x = center[0] + (R+r) * np.cos(t) - r * np.cos((R+r)*t/r)
            y = center[1] + (R+r) * np.sin(t) - r * np.sin((R+r)*t/r)
        elif shape == "hypotrochoid":
            x = center[0] + (R-r) * np.cos(t) + d * np.cos((R-r)*t/r)
            y = center[1] + (R-r) * np.sin(t) - d * np.sin((R-r)*t/r)
        elif shape == "epitrochoid":
            x = center[0] + (R+r) * np.cos(t) - d * np.cos((R+r)*t/r)
            y = center[1] + (R+r) * np.sin(t) - d * np.sin((R+r)*t/r)
            
        plt.plot(x, y, color=color_hex, linewidth=3*size_factor)
    
    elif shape in ["spherical cap", "spherical wedge", "spherical lune"]:
        if shape == "spherical cap":
            # Approximate with an ellipse
            ellipse = patches.Ellipse(center, base_size*2, base_size, color=color_hex)
            ax.add_patch(ellipse)
        elif shape == "spherical wedge":
            # Approximate with a sector
            theta = np.linspace(0, np.pi/3, 100)
            x = center[0] + base_size * np.cos(theta)
            y = center[1] + base_size * np.sin(theta)
            x = np.append(x, [center[0]])
            y = np.append(y, [center[1]])
            sector = plt.Polygon(list(zip(x, y)), color=color_hex)
            ax.add_patch(sector)
        elif shape == "spherical lune":
            # Approximate with a lune
            circle1 = plt.Circle((0.5 - base_size*0.3, 0.5), base_size, color=color_hex, alpha=0.7)
            circle2 = plt.Circle((0.5 + base_size*0.3, 0.5), base_size, color=color_hex, alpha=0.7)
            ax.add_patch(circle1)
            ax.add_patch(circle2)
    
    elif shape == "superellipse":
        t = np.linspace(0, 2*np.pi, 100)
        n = 4  # Controls the shape (higher n = more square-like)
        a, b = base_size * 1.5, base_size * 1.5
        x = center[0] + a * np.sign(np.cos(t)) * np.abs(np.cos(t))**(2/n)
        y = center[1] + b * np.sign(np.sin(t)) * np.abs(np.sin(t))**(2/n)
        superellipse = plt.Polygon(list(zip(x, y)), color=color_hex)
        ax.add_patch(superellipse)
    
    else:
        # Default for any unhandled shapes
        text = f"{shape}"
        ax.text(0.5, 0.5, text, fontsize=16*size_factor, ha='center', va='center', color=color_hex)
    
    return color_name, size_factor * 100  # Return color name and size percentage

def generate_caption(shape, color, size_percent):
    """Generate diverse captions in random order."""
    size_description = get_size_description(size_percent)
    
    # Create the caption components
    components = [
        f"a {shape}",
        f"a {color} shape",
        f"a {size_description} object",
        f"a shape that is {size_percent:.1f}% of maximum size"
    ]
    
    # Randomly select and order the components
    selected_components = random.sample(components, k=random.randint(2, 4))
    
    # Join the components into a caption
    caption = " ".join(selected_components)
    
    # Sometimes completely randomize the order of words
    if random.random() < 0.25:
        all_words = caption.split()
        random.shuffle(all_words)
        caption = " ".join(all_words)
    
    alternatives = [
        f"This image shows {caption}.",
        f"An illustration containing {caption}.",
        f"A {color} {size_description} {shape}.",
        f"The picture depicts {caption}.",
        f"{caption.capitalize()}.",
        f"A {size_description} {color} {shape} at {size_percent:.1f}% size.",
        f"The image features {caption}.",
        f"{size_description.capitalize()} {color} {shape}.",
        f"{shape.capitalize()} ({color}, {size_percent:.1f}% size).",
        f"{color.capitalize()} {shape} of {size_description} size."
    ]

    # Add more diversity to the captions
    alternatives.extend([
        # Simple alternatives
        f"A {shape}.",
        f"{color.capitalize()} {shape}.",
        f"{caption}.",
        f"Image: {shape}.",
        f"A geometric figure.",
        
        # Descriptive alternatives
        f"A finely detailed {color} {shape} rendered with precision at {size_percent:.1f}% of standard dimensions.",
        f"The illustration presents a meticulously crafted {color} {shape}, notable for its {size_description} proportions and clean lines.",
        f"This high-quality visual representation features a {color} {shape} with dimensions calibrated to {size_percent:.1f}% of the reference size.",
        f"An expertly rendered {size_description} {color} {shape} displayed with attention to geometric accuracy.",
        f"The image showcases a precisely defined {color} {shape} with carefully maintained proportions at {size_percent:.1f}% scale.",
        
        # Minimal description
        f"{shape}.",
        f"{color}.",
        f"Figure.",
        f"Geometric.",
        f"Visual.",
        
        # All variables
        f"A {size_description} {color} {shape} rendered at {size_percent:.1f}% scale depicting {caption}.",
        f"This {size_description} {color} {shape} ({size_percent:.1f}%) represents {caption}.",
        f"Visual element: {size_description} {color} {shape}, {size_percent:.1f}%, {caption}.",
        f"{caption}: {size_description} {color} {shape} at {size_percent:.1f}% relative size.",
        f"{color.capitalize()} {size_description} {shape} ({size_percent:.1f}%) displaying {caption}.",
        
        # No variables (completely generic)
        f"A geometric shape.",
        f"A visual element.",
        f"A colored shape.",
        f"A geometric figure.",
        f"A visual representation.",
        
        # Technical focus
        f"Object class: {shape}; color: {color}; size modifier: {size_percent:.1f}%.",
        f"Visual asset: type={shape}, color={color}, size={size_percent:.1f}%, description='{caption}'.",
        f"Element[{shape}]: color={color}, scale={size_percent:.1f}%, desc='{caption}'.",
        f"Geometric representation (type: {shape}, color: {color}, scale: {size_percent:.1f}%).",
        f"Visual.render({shape}, {color}, {size_percent:.1f}%, '{caption}').",
        
        # Creative alternatives
        f"Behold, a {color} {shape} of {size_description} dimensions!",
        f"Gaze upon this {size_description} {color} {shape}, a testament to geometric elegance.",
        f"A wild {color} {shape} appears! It seems to be {size_description}.",
        f"The canvas reveals a {color} {shape}, sized at a modest {size_percent:.1f}% of potential.",
        f"Lo! A {color} {shape} of {size_description} proportions graces your vision.",
        
        # Mixed variable usage
        f"A {color} figure at {size_percent:.1f}% scale.",
        f"{size_description.capitalize()} {shape}.",
        f"{color.capitalize()} geometric element depicting {caption}.",
        f"Shape type: {shape}; Size: {size_description}.",
        f"This {color} element represents {caption}.",
        
        # Academic style
        f"The figure presents a {color} {shape} (n={size_percent:.1f}%, p<0.05) consistent with {caption}.",
        f"Fig. 1: {color.capitalize()} {shape} at {size_percent:.1f}% relative scale.",
        f"A {size_description} {color} {shape}, characteristic of category '{caption}'.",
        f"Visual stimulus: {color} {shape} (scale factor: {size_percent:.1f}).",
        f"Specimen: {color} {shape}, scaled to {size_percent:.1f}% of reference dimensions.",
        
        # Alternative sentence structures
        f"What you see is a {color} {shape}.",
        f"There exists a {size_description} {color} {shape} in this image.",
        f"Shown here: {color} {shape}, {size_description}.",
        f"Present in the visual field: one {color} {shape}.",
        f"Contained within: a {size_description} {color} {shape}.",
        
        # Using only some variables in different combinations
        f"A {color} form.",
        f"The {size_description} element.",
        f"{size_percent:.1f}% sized object.",
        f"{caption} represented by a {shape}.",
        f"{shape.capitalize()} expressing {caption}."
    ])
    
    return random.choice(alternatives)

def get_folder_name(image_index, images_per_folder=5000):
    """Generate folder name based on image index."""
    folder_number = (image_index // images_per_folder) + 1
    return f"dataset/images_{folder_number:03d}"

def write_to_csv(file_path, data):
    """Write data to a CSV file."""
    with open(file_path, mode='w', newline='') as file:
        writer = csv.writer(file)
        writer.writerow(["file_name", "caption"])
        for folder_name, file_name, caption in data:
            writer.writerow([file_name, caption])

def generate_single_image_process(i, img_size=(512, 512), images_per_folder=5000):
    """Process-safe version of image generation."""
    # Each process needs its own matplotlib backend setup
    import matplotlib
    matplotlib.use('Agg')
    import matplotlib.pyplot as plt
    
    shape = random.choice(shapes)
    size_factor = random.uniform(0.1, 1.0)  # Random size from 10% to 100%
    
    folder_name = get_folder_name(i, images_per_folder)
    os.makedirs(folder_name, exist_ok=True)
    
    file_name = f"image_{i+1}.png"
    file_path = os.path.join(folder_name, file_name)
    
    # Calculate figure size in inches based on desired pixel dimensions and DPI
    dpi = 100  # Standard DPI for matplotlib
    fig_size = (img_size[0] / dpi, img_size[1] / dpi)  # Convert pixels to inches
    
    fig, ax = plt.subplots(figsize=fig_size, dpi=dpi)
    ax.set_xlim(0, 1)
    ax.set_ylim(0, 1)
    ax.set_xticks([])
    ax.set_yticks([])
    ax.set_frame_on(False)
    
    color_name, size_percent = draw_shape(shape, ax, size_factor)
    caption = generate_caption(shape, color_name, size_percent)
    
    # Save the figure with the exact dimensions
    plt.savefig(file_path, dpi=dpi, bbox_inches=None, pad_inches=0)  # Disable bbox_inches and pad_inches
    plt.close(fig)
    
    # Write the caption to a CSV file in the same folder
    csv_path = os.path.join(folder_name, "metadata.csv")
    with open(csv_path, mode='a', newline='') as file:
        writer = csv.writer(file)
        # Write header only if the file is empty or doesn't exist
        if not os.path.exists(csv_path) or os.path.getsize(csv_path) == 0:
            writer.writerow(["file_name", "caption"])
        writer.writerow([file_name, caption])
    
    return folder_name, file_name, caption

def generate_images_multiprocess(num_images=10, img_size=(512, 512), max_workers=None, images_per_folder=5000):
    """Generate images using multiple processes instead of threads."""
    # Create dataset directory structure
    os.makedirs("dataset", exist_ok=True)
    
    start_time = time.time()
    
    # Determine number of workers if not specified
    if max_workers is None:
        max_workers = mp.cpu_count()
    
    # Create a pool of workers
    with mp.Pool(processes=max_workers) as pool:
        # Create a partial function with fixed img_size
        worker_func = partial(generate_single_image_process, img_size=img_size, images_per_folder=images_per_folder)
        
        # Map the function to the range of image indices
        results = list(tqdm(
            pool.imap(worker_func, range(num_images)),
            total=num_images,
            desc="Generating Images"
        ))
    
    end_time = time.time()
    
    print(f"Generated {num_images} images in {end_time - start_time:.2f} seconds")
    print(f"Size: {len(results)} images")

if __name__ == "__main__":
    # You can adjust the number of workers based on your CPU
    # If left as None, it will use the default (typically the number of CPU cores)
    generate_images_multiprocess(num_images=100000, max_workers=10, img_size=(512, 512), images_per_folder=5000)