draw/trace_data_new.py at master · numberlearning/draw · GitHub

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
import sys
import numpy as np
import random
import matplotlib as mpl
import matplotlib.pyplot as plt

#np.set_printoptions(threshold=np.nan)


def get_my_teacher():
    min_edge = 7
    max_edge = 7
    min_blobs = 1
    max_blobs = 15


    c = min_blobs
    total = 0
    while (c < max_blobs + 1):
       if c == 0:
           total = total + 1000
           c = c + 1
           continue
       cN = 1000#int(10/c)
       total = total + cN
       c = c + 1

    nOf_nBlob_imgs = 4
    img_height = 40
    img_width = 200
    blob_spacing = 7
    edge_vertical_space = 10
    edge_horizontal_space = 10
    img_size = img_height * img_width   #row x columns
    n_labels = max_blobs  #features in a label
    all_traces = []
    train = np.zeros([total, img_size])   #training imgs
    label = np.zeros([total, n_labels])   #labels
    trace = []          #list(imgs) of set(traces for each img) of tuples(lx, ly)
    img_count = 0
    length = 0

    #create images for each number of blobs
    for n in range(min_blobs, max_blobs+1):
        num_blobs = n       #1 to 9
        num_imgs = 1000#int(10/n)      #number of images to generate for this number of blob

        #generate individual image
        for x in range(num_imgs):
            img = np.zeros(img_size)
            count = 0
            used = np.zeros([num_blobs, 4]) # x_coor, y_coor, width, height for each blob
            list_traces = []  #list of traces
            lxly_list = []   #x,y coordinate tuples of blobs in the image

            #create info of each image
            while count < num_blobs:
                height = random.randint(min_edge, max_edge)
                width = random.randint(min_edge, max_edge)
                lx = random.randint(edge_vertical_space, img_width - edge_horizontal_space - width)
                ly = random.randint(edge_vertical_space, img_height - edge_vertical_space - height)
                h = height
                w = width

                index = 0

                #find valid positions for the blobs
                while index < count:
                    if lx+width+1 <= used[index, 0] or used[index, 0]+1+used[index,2] <= lx or used[index, 1]+1+used[index,3] <= ly or ly+height+1<=used[index,1]:
                        if index == count-1:
                            position = count - 1
                            #while position >= 0:
                            #    hori1 =  (lx >= used[position, 0] - width - 6) and (lx <= used[position, 0] - width - 1)
                            #    hori2 =  (lx >= used[position, 0] + used[position,2] + 1) and (lx <= used[position, 0] + used[position,2] + 6)
                            #    vert1 =  (ly >= used[position, 1] - height - 3) and (ly <= used[count-1, 1] - height - 1)
                            #    vert2 =  (ly >= used[position, 1] + used[position, 3] + 1) and (ly <= used[position, 1] + used[position, 3] + 3)
                            #    if (hori1 or hori2) and (vert1 or vert2):
                            #        index = index + 1
                            #        break
                            #    position = position - 1
                            #if index == count:
                            #    lx = random.randint(edge_horizontal_space, img_width - edge_horizontal_space - width)
                            #    ly = random.randint(edge_vertical_space, img_height - edge_vertical_space - height)
                            #    index = 0
                            while position >= 0:
                                distance = (lx - used[position, 0])**2 + (ly - used[position, 1])**2
                                if distance <= 338:
                                    index = index + 1
                                    break
                                position = position - 1
                            if index < count:
                                lx = random.randint(edge_horizontal_space, img_width - edge_horizontal_space - width)
                                ly = random.randint(edge_vertical_space, img_height - edge_vertical_space - height)
                                index = 0
                        else:
                            index = index + 1
                    else:
			#left_bound1 = used[count, 0] - width - 5
	                #right_bound1 = used[
			#right_bound2 = used[count, 0] + used[count, 2] + 5
			#top_bound = used[count, 1] + used[count, 3] + 5
			#bottom_bound = used[count, 1] - height - 5
			#left = left_bound if left_bound > edge_horizontal_space else edge_horizontal_space
                        lx = random.randint(edge_horizontal_space, img_width - edge_horizontal_space - width)
                        ly = random.randint(edge_vertical_space, img_height - edge_vertical_space - height)
                        index = 0

                used[index, 0] = lx
                used[index, 1] = ly
                lxly_list.append((lx + int(width/2), ly + int(height/2)))
               # print(lxly_list)
                used[index, 2] = width
                used[index, 3] = height

                for p in range(ly, ly+height):
                    for q in range(lx, lx+width):
                        img[p*img_width+q] = 255

                count = count + 1
                #intensity = intensity + int(255/num_blobs)

            train[img_count] = img
            label[img_count, num_blobs-1] = 1
            img_count += 1
            img = img.reshape(img_height, img_width)
            #plt.imshow(img, interpolation="nearest", origin="upper")
            #plt.colorbar()
            #plt.title(label[img_count - 1])
            #plt.show()

            #create traces for each image
            for t in range(num_blobs):
                #trace_current = []
                l1 = lxly_list
                if t % 2 == 0:
                    l1.sort(key=lambda tup:tup[0])
                #    l1.sort(key=lambda tup:tup[1])
                if t % 2 == 1:
                    l1.sort(key=lambda tup:tup[0], reverse=True)
                #if t % 4 == 2:
                #    l1.sort(key=lambda tup:tup[0])
                #    l1.sort(key=lambda tup:tup[1], reverse=True)
                #if t % 4 == 3:
                #    l1.sort(key=lambda tup:tup[0], reverse=True)
                #    l1.sort(key=lambda tup:tup[1], reverse=True)
                #for index in range(num_blobs):
                    #trace_current.append(l1[index])
                    minimum = 10000000000
                    for p in range(1,num_blobs-index):
                        diff = (l1[index+p][0]-l1[index][0])**2 + (l1[index+p][1]-l1[index][1])**2
                        if diff < minimum:
                            minimum = diff
                            temp = l1[index+1]
                            l1[index+1] = l1[index+p]
                            l1[index+p] = temp
                        p = p + 1
                    index = index+1
                v =int(255 / num_blobs)
                im = np.zeros((img_height, img_width))
                for j in range(num_blobs):
                    for y in range(-2, 3):
                        for x in range(-2, 3):
                            im[l1[j][1]+x][l1[j][0]+y] = v
                    v = v + int(255 / num_blobs)
                trace_vector = np.zeros(img_size)
                v = int(255/num_blobs)
                for j in range(num_blobs):
                    for y in range(-2, 3):
                        for x in range(-2, 3):
                            trace_vector[(l1[j][1]+y)*img_width+(l1[j][0]+x)]=v
                            #print((l1[j][0]+y)*img_width + (l1[j][1]+x))
                    v = v + int(255/num_blobs)
                imggg = trace_vector.reshape(img_height, img_width)
                #print(im)
            #    plt.imshow(im, interpolation="nearest", origin="upper")
            #    plt.colorbar()
            #    plt.title(label[img_count - 1])
            #    plt.show()
                list_traces.append(l1)
                length = length + 1
                all_traces.append(trace_vector)

            trace.append(list_traces)
            #print(trace)

    return all_traces, train, label, trace, length