-
Notifications
You must be signed in to change notification settings - Fork 0
/
test_model.py
149 lines (104 loc) · 4.03 KB
/
test_model.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
147
148
149
import matplotlib
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import glob
import os
import torch
#from IPython.display import clear_output
from skimage.io import imread
from skimage import transform as tf
from skimage.transform import resize
import cv2
from PIL import Image
import random
import scipy
#from sklearn.preprocessing import normalize
from os.path import basename
#import regex as re
import os
import torchvision
import torch
from torch.autograd import Variable
import torch.nn as nn
import torch.optim as optim
from torch.nn import Linear, GRU, Conv2d, Dropout, MaxPool2d, BatchNorm1d
from torch.nn.functional import relu, elu, relu6, sigmoid, tanh, softmax
from torch.utils.data import DataLoader, random_split
from augmentation.image_processing import*
model = torch.load("/zhome/ed/a/183709/saved_model_dice_deepCE_smol_W_col.pth")
# imports the image processing packets
#from import_data import*
#image_paths = image_path()
from unet.Network import *
# imports the test images
#from batch_loader import test_images
from sklearn import preprocessing
#test_one_hot_, test_image_, test_image_standard_, test_image_standard_hot_ = test_images()
drive_path = r"/zhome/ed/a/183709/data/data/test/*.*"
image_paths = glob.glob(drive_path)
test_image_ = [np_transform_bgr(np.load(i)[0:3, :, 🙂) for i in image_paths]
test_one_hot_ = [one_hot_image(np.load(i)) for i in image_paths]
test_image_standard_hot_show = [np.expand_dims(np.load(i)[3,:,:], axis = 0) for i in image_paths]
test_image_standard_hot_ = [one_trans(x) for x in test_image_standard_hot_show]
test_image_standard_ = [np.load(i)[0:3,:,:] for i in image_paths] #for non normal
#test_image_standard_ = [np.expand_dims(preprocessing.normalize(rgb_grey(np.load(i))), axis = 0) for i in image_paths]
#test_image_standard_ = [np.expand_dims(rgb_grey(np.load(i)), axis = 0) for i in image_paths] #for non normal
test_image_standard = test_image_standard_
test_image_standard_hot = test_image_standard_hot_
#test_image_standard = [np.load(i)[0:3,:,:] for i in test_image_standard_ ]
#test_image_standard_hot = [(np.load(i)[3,:,:]) for i in test_image_standard_hot_ ]
#test_image_standard_hot = [one_trans(x) for x in test_image_standard_hot]
X_t = torch.from_numpy(np.array(test_image_standard, dtype = 'float32'))
X_t= X_t.to(device='cuda')
Y_t = torch.from_numpy(np.array(test_image_standard_hot, dtype ='float32'))
Y_t = Y_t.to(device = 'cuda')
# imports the network
def accuracy(ys, ts):
predictions = torch.max(ys, 1)[1]
correct_prediction = torch.eq(predictions, ts)
return torch.mean(correct_prediction.float())
print(X_t.shape)
device = "cuda" if torch.cuda.is_available() else "cpu"
Net = U_Net_Model()
Net = torchvision.models.segmentation.deeplabv3_resnet101(num_classes = 9)
Net.backbone.conv1 = nn.Conv2d(3, 64, kernel_size=7, stride=2, padding=3, bias=False)
if torch.cuda.is_available():
Net.cuda()
Net.load_state_dict(model)
Net.eval()
def nine_to_one(x):
img = np.zeros((256,256))
for i in range(len(x)):
img[:,:] += x[i,:,:]*i
return img
select_test = np.arange(30)
from pytorch_toolbelt.losses import DiceLoss
criterion1 = DiceLoss( mode = 'multilabel')
count = 0
avg_dice = 0
for i,x in enumerate(select_test):
test = Net(X_t[x:x+1])['out']
acc = criterion1(test, Y_t[x:x+1])
acc = acc.item()
#_, test = torch.max(test.data, 1)
test = torch.max(test, 1)[1]
test= test.data.cpu().detach().numpy()
test = np.array(test)[0]
f, axarr = plt.subplots(1,3)
axarr[0].imshow(test)
axarr[0].title.set_text("Prediction")
axarr[0].axis('off')
axarr[1].imshow(test_image_standard_hot_show[x][0])
axarr[1].title.set_text("Target")
axarr[1].axis('off')
axarr[2].imshow(test_image_[x])
axarr[2].title.set_text("Original image")
axarr[2].axis('off')
f.suptitle("Dice Score: " + str(1 - acc)[:4])
f.tight_layout()
f.subplots_adjust(top=1.35)
plt.savefig('/zhome/ed/a/183709/lol/comparison'+str(x) + '.png')
plt.show()
avg_dice += acc/30
print(avg_dice)