-
Notifications
You must be signed in to change notification settings - Fork 2
/
render_VisualAtom.py
130 lines (106 loc) · 5.64 KB
/
render_VisualAtom.py
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
import math
from PIL import Image, ImageDraw
import random
import noise
import os
import argparse
########################################################################################
def conf():
parser = argparse.ArgumentParser()
parser.add_argument("--save_root", default="./VisualAtom_dataset",type=str, help="path to image file save directory")
parser.add_argument("--numof_classes", default=1000, type=int, help="Visual Atom category number")
parser.add_argument("--numof_instances", default=1000, type=int, help="Visual Atom instance number")
parser.add_argument("--start_class", default=0, type=int, help="Visual Atom start class number")
parser.add_argument("--image_size", default=512, type=int)
parser.add_argument("--numof_thread", default=1, type=int, help="")
parser.add_argument("--thread_num", default=0, type=int, help="")
# Category parameter setting
parser.add_argument("--vertex_num_max", default=1000, type=int, help="")
parser.add_argument("--vertex_num_min", default=200, type=int, help="")
parser.add_argument("--perlin_min", default=2, type=int, help="")
parser.add_argument("--line_width", default=0.01, type=float, help="")
parser.add_argument("--radius_min", default=0, type=int, help="")
parser.add_argument("--line_num_min", default=1, type=int, help="")
parser.add_argument("--line_num_max", default=200, type=int, help="")
parser.add_argument("--oval_rate", default=2, type=int, help="")
parser.add_argument("--start_pos", default=512, type=int, help="")
parser.add_argument("--nami_1_min", default=0, type=int, help="")
parser.add_argument("--nami_2_min", default=0, type=int, help="")
parser.add_argument("--nami_1_max", default=20, type=int, help="")
parser.add_argument("--nami_2_max", default=20, type=int, help="")
# Display on screen
parser.add_argument("--display", action='store_true', help="Display the generated images")
args = parser.parse_args()
return args
args = conf()
vertex_x = []
vertex_y = []
point_x = []
point_y = []
Noise_x = []
Noise_y = []
im = []
vertex_number = 3
nami1 = 0
nami2 = 0
random.seed(args.thread_num + 1)
class_per_thread = args.numof_classes / args.numof_thread
cat_start = args.thread_num * int(class_per_thread) + args.start_class
cat_finish = cat_start + int(class_per_thread)
for cat in range(int(cat_start), int(cat_finish)):
if not os.path.exists(os.path.join(args.save_root, "%05d" % cat)):
os.makedirs(os.path.join(args.save_root, "%05d" % cat))
vertex_number = random.randint(args.vertex_num_min, args.vertex_num_max)
line_draw_num = random.randint(args.line_num_min, args.line_num_max)
perlin_noise_coefficient = random.uniform(args.perlin_min, (args.perlin_min + 4))
line_width = random.uniform(0.0, args.line_width)
start_rad = random.randint(args.radius_min, args.radius_min + 50)
nami1 = random.randint(args.nami_1_min, args.nami_1_max)
while True:
nami2 = random.randint(args.nami_2_min, args.nami_2_max)
if (nami1 != nami2):
break
elif (nami1 == 0):
break
oval_rate_x = random.uniform(1, args.oval_rate)
oval_rate_y = random.uniform(1, args.oval_rate)
start_pos_h = (args.image_size + random.randint(-1 * args.start_pos, args.start_pos)) / 2
start_pos_w = (args.image_size + random.randint(-1 * args.start_pos, args.start_pos)) / 2
vertex_x.clear()
vertex_y.clear()
im.clear()
for k2 in range(args.numof_instances):
im.append(Image.new('RGB', (args.image_size, args.image_size), (0, 0, 0)))
draw = ImageDraw.Draw(im[k2])
angle = (math.pi * 2) / vertex_number
for vertex in range(vertex_number):
vertex_x.append(math.cos(angle * vertex) * start_rad * oval_rate_x + start_pos_w)
vertex_y.append(math.sin(angle * vertex) * start_rad * oval_rate_y + start_pos_h)
vertex_x.append(vertex_x[0])
vertex_y.append(vertex_y[0])
for line_draw in range(line_draw_num):
gray = random.randint(0, 255)
Noise_x.clear()
Noise_y.clear()
for vertex in range(vertex_number):
Noise_x.append(random.uniform(0 , 10000))
Noise_x[vertex] = noise.pnoise1(Noise_x[vertex]) * perlin_noise_coefficient - perlin_noise_coefficient - 0.5*math.sin(angle * vertex * nami1) - 0.5*math.sin(angle * vertex * nami2)
for vertex in range(vertex_number):
Noise_y.append(random.uniform(0 , 10000))
Noise_y[vertex] = noise.pnoise1(Noise_y[vertex]) * perlin_noise_coefficient - perlin_noise_coefficient - 0.5*math.sin(angle * vertex * nami1) - 0.5*math.sin(angle * vertex * nami2)
for vertex in range(vertex_number):
vertex_x[vertex] -= math.cos(angle * vertex) * (Noise_x[vertex] - line_width)
vertex_y[vertex] -= math.sin(angle * vertex) * (Noise_y[vertex] - line_width)
vertex_x[vertex_number] = vertex_x[0]
vertex_y[vertex_number] = vertex_y[0]
for i in range(vertex_number):
draw.line((vertex_x[i], vertex_y[i], vertex_x[i + 1], vertex_y[i + 1]) , fill = (gray, gray, gray), width = 1)
if not args.display:
im[k2].save(args.save_root + "/%05d/vertex_%04d_instance_%04d.png" % (cat, vertex_number, k2), quality = 95)
else:
im[k2].show()
vertex_x.clear()
vertex_y.clear()
start_pos_h = (args.image_size + random.randint(-1 * args.start_pos, args.start_pos)) / 2
start_pos_w = (args.image_size + random.randint(-1 * args.start_pos, args.start_pos)) / 2
print('Gerated Category:' + str(cat))