/
patchgen_test.py
146 lines (93 loc) · 4.21 KB
/
patchgen_test.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
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
import gp
import glob
import mahotas as mh
import numpy as np
import os
import time
def read_section(path, z, verbose=True, withNP=True):
'''
'''
image = sorted(glob.glob(os.path.join(path, 'image', '*_'+str(z).zfill(9)+'_image.png')))
mask = sorted(glob.glob(os.path.join(path, 'mask', 'z'+str(z).zfill(8)+'.png')))
gold = sorted(glob.glob(os.path.join(path, 'gold', 'z'+str(z).zfill(8)+'.png')))
if withNP:
# load from neuroproof
rhoana = sorted(glob.glob(os.path.join(path, 'neuroproof', '*_'+str(z).zfill(9)+'_neuroproof.png')))
else:
# load from segmentation
rhoana = sorted(glob.glob(os.path.join(path, 'segmentation', '*_'+str(z).zfill(9)+'_segmentation.png')))
prob = sorted(glob.glob(os.path.join(path, 'prob', str(z).zfill(4)+'.png')))
if verbose:
print 'Loading', os.path.basename(image[0])
image = mh.imread(image[0])
mask = mh.imread(mask[0]).astype(np.bool)
gold = mh.imread(gold[0])
rhoana = mh.imread(rhoana[0])
prob = mh.imread(prob[0])
#convert ids from rgb to single channel
rhoana_single = np.zeros((rhoana.shape[0], rhoana.shape[1]), dtype=np.uint64)
rhoana_single[:, :] = rhoana[:,:,0]*256*256 + rhoana[:,:,1]*256 + rhoana[:,:,2]
gold_single = np.zeros((gold.shape[0], gold.shape[1]), dtype=np.uint64)
gold_single[:, :] = gold[:,:,0]*256*256 + gold[:,:,1]*256 + gold[:,:,2]
# # relabel the segmentations
# gold_single = Util.relabel(gold_single)
# rhoana_single = Util.relabel(rhoana_single)
#mask the rhoana output
rhoana_single[mask==0] = 0
return image, prob, mask, gold_single, rhoana_single
def generate_patches(start_slice, end_slice, withNP):
patch_index = 0
all_patches_count = 0
patch_list = []
all_error_patches = []
all_correct_patches = []
for z in range(start_slice, end_slice):
t0 = time.time()
print 'working on slice', z
input_image, input_prob, input_mask, input_gold, input_rhoana = read_section('/n/regal/pfister_lab/haehn/FINAL/cylinder/',z, False, withNP)
error_patches, patches = gp.Patch.patchify_maxoverlap(input_image, input_prob, input_mask, input_rhoana, input_gold, sample_rate=1)
print 'Generated', len(error_patches), 'split error patches and', len(patches), ' correct patches in', time.time()-t0, 'seconds.'
patch_list.append(patches)
all_error_patches += error_patches
all_correct_patches += patches
NO_PATCHES = len(all_error_patches) + len(all_correct_patches)
print 'We have a total of',NO_PATCHES,'patches.'
print 'Errors:',len(all_error_patches)
print 'Correct:',len(all_correct_patches)
PATCH_BYTES = 75*75
P_SIZE = (NO_PATCHES, 4, 75,75) # rather than raveled right now
p_rgba = np.zeros(P_SIZE, dtype=np.float32)
p_target = np.zeros(NO_PATCHES)
i = 0
for p in all_error_patches:
p_rgba[i][0] = p['image']
p_rgba[i][1] = 1. - p['prob'] ### INVERT PROB
p_rgba[i][2] = p['merged_array']
p_rgba[i][3] = p['larger_border_overlap']
p_target[i] = 1 # <--- important
i += 1
for p in all_correct_patches:
p_rgba[i][0] = p['image']
p_rgba[i][1] = 1. - p['prob'] ### INVERT PROB
p_rgba[i][2] = p['merged_array']
p_rgba[i][3] = p['larger_border_overlap']
p_target[i] = 0 # <--- important
i+=1
return p_rgba, p_target
def shuffle_in_unison_inplace(a, b):
assert len(a) == len(b)
p = np.random.permutation(len(a))
return a[p], b[p]
def run(PATCH_PATH, start_slice, end_slice, filename, withNP):
if not os.path.exists(PATCH_PATH):
os.makedirs(PATCH_PATH)
p = generate_patches(start_slice, end_slice, withNP)
shuffled = shuffle_in_unison_inplace(p[0],
p[1]
)
print 'saving..'
np.savez(PATCH_PATH+filename+'.npz', rgba=shuffled[0])
np.savez(PATCH_PATH+filename+'_targets.npz', targets=shuffled[1])
print 'Done!'
run('/n/regal/pfister_lab/haehn/FINAL/IPMLB/', 250, 300, 'test', True)
#run('/n/regal/pfister_lab/haehn/FINAL/IPMLB_before_NP', 10, 20, 'train', False)