-
Notifications
You must be signed in to change notification settings - Fork 1
/
BCI_TrainingD.py
259 lines (246 loc) · 8.22 KB
/
BCI_TrainingD.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
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
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
import csv
import argparse
import os
from sklearn import metrics
from sklearn.metrics import roc_auc_score
from matplotlib import pyplot
from sklearn.metrics import f1_score
from sklearn.metrics import auc
from sklearn.metrics import confusion_matrix
import seaborn as sns
#provide inputs
parser = argparse.ArgumentParser()
parser.add_argument('--det',type=str) #give path to detection files
parser.add_argument('--gt',type=str) #give path to ground truth files
args = parser.parse_args()
#graph style
sns.set_style('whitegrid')
image_names= []
driver_ground_truth= []
front_ground_truth= []
back_ground_truth= []
# retrieve ground truth
for root, dirs, files in os.walk(args.gt):
if not files:
continue
prefix = os.path.basename(root)
files.sort()
for f in files:
if f.endswith('.txt'):
image_names.append(f)
with open(os.path.join(root, f)) as txt_file:
driver_found= False
front_found= False
back_found= False
for line in txt_file:
type= line.split()[0]
if type=='0':
driver_found = True
elif type=='1':
front_found = True
elif type=='2':
back_found = True
if driver_found==False:
driver_ground_truth.append(0)
else:
driver_ground_truth.append(1)
if front_found==False:
front_ground_truth.append(0)
else:
front_ground_truth.append(1)
if back_found==False:
back_ground_truth.append(0)
else:
back_ground_truth.append(1)
#prediction of driver in images
driver_pred_res= []
vertical_limit= 240
horizontal_limit= 360
bbox_area_limit= 30000
i= 0
for root, dirs, files in os.walk(args.det):
if not files:
continue
prefix = os.path.basename(root)
files.sort()
for f in files:
if f.endswith('.txt'):
if os.path.splitext(f)[0]==os.path.splitext(image_names[i])[0]:
with open(os.path.join(root, f)) as txt_file:
prob= []
for line in txt_file:
type= line.split()[0]
xmin = float(line.split()[2])
ymin = float(line.split()[3])
xmax = float(line.split()[4])
ymax = float(line.split()[5])
conf = float(line.split()[1])
bbox_area= (xmax-xmin)*(ymax-ymin)
x= (xmin+xmax)/2
y= (ymin+ymax)/2
if bbox_area>bbox_area_limit and x>horizontal_limit and y>vertical_limit:
prob.append(conf)
temp = 1
for j in prob:
temp *= (1-j)
final_prob= (1-temp)
driver_pred_res.append(final_prob)
i+=1
#prediction of front passenger in images
front_pred_res= []
vertical_limit= 240
horizontal_limit= 360
bbox_area_limit= 30000
i= 0
for root, dirs, files in os.walk(args.det):
if not files:
continue
prefix = os.path.basename(root)
files.sort()
for f in files:
if f.endswith('.txt'):
if os.path.splitext(f)[0]==os.path.splitext(image_names[i])[0]:
with open(os.path.join(root, f)) as txt_file:
prob= []
for line in txt_file:
type= line.split()[0]
xmin = float(line.split()[2])
ymin = float(line.split()[3])
xmax = float(line.split()[4])
ymax = float(line.split()[5])
conf = float(line.split()[1])
bbox_area= (xmax-xmin)*(ymax-ymin)
x= (xmin+xmax)/2
y= (ymin+ymax)/2
if bbox_area>bbox_area_limit and x<horizontal_limit and y>vertical_limit:
prob.append(conf)
temp = 1
for j in prob:
temp *= (1-j)
final_prob= (1-temp)
front_pred_res.append(final_prob)
i+=1
#prediction of back passenger in images
back_pred_res= []
horizontal_limit1= 250
horizontal_limit2= 720-250
bbox_area_limit= 0
i= 0
for root, dirs, files in os.walk(args.det):
if not files:
continue
prefix = os.path.basename(root)
files.sort()
for f in files:
if f.endswith('.txt'):
if os.path.splitext(f)[0]==os.path.splitext(image_names[i])[0]:
with open(os.path.join(root, f)) as txt_file:
prob= []
for line in txt_file:
type= line.split()[0]
xmin = float(line.split()[2])
ymin = float(line.split()[3])
xmax = float(line.split()[4])
ymax = float(line.split()[5])
conf = float(line.split()[1])
bbox_area= (xmax-xmin)*(ymax-ymin)
x= (xmin+xmax)/2
y= (ymin+ymax)/2
if bbox_area>bbox_area_limit and x>horizontal_limit1 and x<horizontal_limit2:
prob.append(conf)
temp = 1
for j in prob:
temp *= (1-j)
final_prob= (1-temp)
back_pred_res.append(final_prob)
i+=1
# confusion matrix for driver
print('driver')
limit= 0.87
yconf= []
for i in driver_pred_res:
if i<limit:
yconf.append(0)
else:
yconf.append(1)
a= confusion_matrix(driver_ground_truth, yconf).ravel()
if len(a)==1:
print("all true positive")
else:
tn, fp, fn, tp = a
acc= (tp+tn)/(tn+ fp+ fn+ tp)
print("True Positives: "+str(tp))
print("True Negatives: "+str(tn))
print("False Positives: "+str(fp))
print("False Negatives: "+str(fn))
print("Accuracy: "+str(acc))
# confusion matrix for front seat passenger
print("front seat")
limit= 0.89
yconf= []
for i in front_pred_res:
if i<limit:
yconf.append(0)
else:
yconf.append(1)
tn, fp, fn, tp = confusion_matrix(front_ground_truth, yconf).ravel()
acc= (tp+tn)/(tn+ fp+ fn+ tp)
print("True Positives: "+str(tp))
print("True Negatives: "+str(tn))
print("False Positives: "+str(fp))
print("False Negatives: "+str(fn))
print("Accuracy: "+str(acc))
# confusion matrix for back seat passenger
print("back seat")
limit= 0.11
yconf= []
for i in back_pred_res:
if i<limit:
yconf.append(0)
else:
yconf.append(1)
tn, fp, fn, tp = confusion_matrix(back_ground_truth, yconf).ravel()
acc= (tp+tn)/(tn+ fp+ fn+ tp)
print("True Positives: "+str(tp))
print("True Negatives: "+str(tn))
print("False Positives: "+str(fp))
print("False Negatives: "+str(fn))
print("Accuracy: "+str(acc))
# # Optional
# # ROC curve
# fpr, tpr, thresholds = metrics.roc_curve(front_ground_truth, front_pred_res, pos_label=1)
# pyplot.plot(fpr, tpr, marker='.', label='Front Seat Passenger')
# fpr, tpr, thresholds = metrics.roc_curve(back_ground_truth, back_pred_res, pos_label=1)
# pyplot.plot(fpr, tpr, marker='.', label='Back Seat Passenger')
# pyplot.xlabel('False Positive Rate')
# pyplot.ylabel('True Positive Rate')
# pyplot.legend()
# pyplot.savefig('ROC.png')
# pyplot.close()
# # Optional
# # calculate AUC
# # example
# roc_auc = roc_auc_score(front_ground_truth, front_pred_res)
# print('ROC curve AUC: %.3f' % roc_auc)
# # Optional
# # Find for which confidence there is the maximum F1 score
# def frange(start, stop, step):
# i = start
# while i < stop:
# yield i
# i += step
# # print(ground_truth)
# max= 0
# yhat= []
# for limit in frange(0.5,1.0,0.001):
# yhat= []
# for i in front_pred_res:
# if i<limit:
# yhat.append(0)
# else:
# yhat.append(1)
# pr_f1 = f1_score(front_ground_truth, yhat)
# if pr_f1>max:
# max= pr_f1
# index= limit
# print('PR curve max F1: %.3f at confidence threshold %.3f' %(max,index))