-
Notifications
You must be signed in to change notification settings - Fork 2
/
CompareModels.py
101 lines (83 loc) · 3.56 KB
/
CompareModels.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
import numpy as np
from keras.models import load_model
import cv2
import matplotlib.pyplot as plt
import os
import sys
import datetime
import platform
import DataSet
from PIL import Image
import CustomMetric
import CustomLoss
def ApplyLaneDetection(Y_hat, XtestImg, generateHeatMap = False):
for i in range(Y_hat.shape[0]):
for j in range(Y_hat.shape[1]):
rvalue = Y_hat[i][j]
if rvalue > 1:
out.write('rvalue = ' + str(rvalue))
XtestImg[i][j] = ( 255, 0, 0)
elif (rvalue >= 0.5):
XtestImg[i][j] = (0, 255, 0)
elif generateHeatMap and (rvalue > 0.25):
XtestImg[i][j] = (np.clip(100 * 2 * rvalue, 0, 100) + 150, 100, 100)
elif generateHeatMap and (rvalue > 0):
XtestImg[i][j] = (100, 100, 150 + np.clip(100 * 4 * rvalue, 0, 100))
def SaveInFolder(id, XtestImg, appendText, out_dir):
im = Image.fromarray(XtestImg)
imgFile = os.path.join(out_dir, str(id) + appendText + '.jpg')
im.save(imgFile)
currentDir = os.path.dirname(os.path.realpath(__file__))
modelDL = load_model(os.path.join(currentDir, 'Model.h5'), custom_objects={'dice_loss': CustomLoss.dice_loss})
modelMSE = load_model(os.path.join(currentDir, 'Model_MSE.h5'))
resultdir = os.getcwd()
if platform.system() == 'Linux':
resultdir = '/var/www/html/results'
mydir = os.path.join(resultdir, datetime.datetime.now().strftime('%Y-%m-%d_%H-%M-%S'))
os.makedirs(mydir)
num_images = 100
#ds = DataSet.TrainDataGenerator(DataSet.dev_set_count, 'dev', 0.2)
ds = DataSet.TrainDataGenerator(num_images, 'test', 0.2)
devSet = next(ds)
XList = devSet[0]
YList = devSet[1]
Y_hat_DL = modelDL.predict(XList)
Y_hat_MSE = modelMSE.predict(XList)
h = XList[0].shape[0] // 2
w = XList[0].shape[1] // 2
hw = '' #' height=' + str(h) + " width=" + str(w)
htmlfile = os.path.join(mydir, 'index.html')
out = open(htmlfile, 'w')
out.write('<html><body><table border=1><tr><td>Ground Truth</td><td>Predicted Using DiceLoss</td><td>Predicted Lane Using MSE probability > 0.5</td><td>MSE HeatMap</td></tr>')
f1_scores = [0] * 9
for ii in range(XList.shape[0]):
for jj in range(9):
f1_scores[jj] = f1_scores[jj] + CustomMetric.compute_f1(YList[ii], Y_hat_DL[ii], (jj + 1) / 10)
tempDL = np.copy(XList[ii])
tempMSE = np.copy(XList[ii])
tempMSE_HM = np.copy(XList[ii])
ApplyLaneDetection(YList[ii], XList[ii])
ApplyLaneDetection(Y_hat_DL[ii], tempDL)
ApplyLaneDetection(Y_hat_MSE[ii], tempMSE)
ApplyLaneDetection(Y_hat_MSE[ii], tempMSE_HM, True)
SaveInFolder(ii, XList[ii], '', mydir)
SaveInFolder(ii, tempDL, '_DL', mydir)
SaveInFolder(ii, tempMSE, '_MSE', mydir)
SaveInFolder(ii, tempMSE_HM, '_MSE_HM', mydir)
out.write('<tr>')
out.write('<td><img src=\'' + str(ii) + '.jpg\'' + hw + '></td>')
out.write('<td><img src=\'' + str(ii) + '_DL.jpg\'' + hw + '></td>')
out.write('<td><img src=\'' + str(ii) + '_MSE.jpg\'' + hw + '></td>')
out.write('<td><img src=\'' + str(ii) + '_MSE_HM.jpg\'' + hw + '></td>')
out.write('<td>' + str(CustomMetric.compute_f1(YList[ii], Y_hat_DL[ii], 0.5)) + '</td>')
out.write('</tr>')
avg_f1_scores = [x / num_images for x in f1_scores]
plt.xlabel('probability threshold')
plt.ylabel('f1 score across dev set')
plt.plot([0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9], avg_f1_scores)
pltfile = os.path.join(mydir, 'plt_f1_score.png')
plt.savefig(pltfile)
plt.clf()
out.write('<tr><td colspan=4><img src=plt_f1_score.png></td></tr>')
out.write('</table></body></html>')
out.close()