forked from ludles/OralScreen
-
Notifications
You must be signed in to change notification settings - Fork 3
/
Copy pathtest.py
156 lines (130 loc) · 5.88 KB
/
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
147
148
149
150
151
152
153
154
155
156
# -*- coding: utf-8 -*-
import csv
w = 80
THRESHOLD = 0.59
file_train = "../VOC2007_old/2007_train.txt"
val_set = ['000024', '000057']
def is_val(line_in_file):
img_path = line_in_file.split()[0]
img_name = img_path.split('/')[-1].split('.')[0]
return (img_name in val_set)
def dist(point1, point2):
dist = 0
for (x1, x2) in zip(point1, point2):
dist += (x1 - x2)**2
return dist**0.5
def train_set():
train_set = []
with open(file_train, 'r') as f:
for line in f.readlines():
img_path = line.split()[0]
img_name = img_path.split('/')[-1].split('.')[0]
if img_name not in val_set:
train_set.append(img_name)
return train_set
train_set = train_set()
def read_data(data_set='val'):
boxes = []
if data_set == 'val':
with open(file_train, 'r') as f:
for line in f.readlines():
if is_val(line):
boxes.extend(line.split()[1:]) # extract the bounding boxes
elif data_set == 'train':
with open(file_train, 'r') as f:
for line in f.readlines():
if not is_val(line):
boxes.extend(line.split()[1:]) # extract the bounding boxes
# ground truth boxes
boxes = [list(map(int, box.split(',')[:-1])) for box in boxes] # convert to int
# ground truth positions
pos = [[(box[0] + box[2]) / 2, (box[1] + box[3]) / 2] for box in boxes]
# read results from csv file
pred_pos = []
for img_name in val_set:
with open('./Results/Results_' + img_name + '.csv') as f:
f_csv = csv.reader(f)
headers = next(f_csv)
_pred_pos = [[float(row[-2]), float(row[-1])] for row in f_csv]
pred_pos.extend(_pred_pos)
return pred_pos, pos
pred_pos, pos = read_data(data_set='val')
#%%
# TP: for each detected nucleus,
# the center of the closest ground truth nucleus is inside the detected window (80, 80)
def evaluate(pred_pos, pos):
tp = 0
tp_fp = len(pred_pos)
tp_fn = len(pos)
# =============================================================================
# # lists for finding failed/false detections
# preds_tp = []
# detected_gt = []
# =============================================================================
for pred_p in pred_pos: # the center of the detected nucleus
# compute distance list between the ground truth nuclei
distance_to_gt = [dist([p[0],p[1]], [pred_p[0],pred_p[1]]) for p in pos]
# the closest ground truth
i_closest_gt = distance_to_gt.index(min(distance_to_gt))
# compute distancce list between the predictions and the closest ground truth
distance_to_pred = [dist([pred[0],pred[1]], [pos[i_closest_gt][0],pos[i_closest_gt][1]]) for pred in pred_pos]
# the clossest prediction to that ground truth
i_closest_pred = distance_to_pred.index(min(distance_to_pred))
# if not match
if (pred_p != pred_pos[i_closest_pred]):
continue
# if the cloest one is inside the detected window
if (pred_p[0] - w/2 <= pos[i_closest_gt][0] <= pred_p[0] + w/2 and
pred_p[1] - w/2 <= pos[i_closest_gt][1] <= pred_p[1] + w/2):
tp += 1
# =============================================================================
# # lists for finding failed/false detections
# preds_tp.append(pred_p)
# detected_gt.append(pos[i_closest_gt])
#
# # 求差集，在前者中但不在后者中
# fp = [list(map(int, pred_p)) for pred_p in pred_pos if pred_p not in preds_tp]
# fn = [list(map(int, p)) for p in pos if p not in detected_gt]
# =============================================================================
precision = tp / tp_fp
recall = tp / tp_fn
f1 = 2 * precision * recall / (precision + recall)
return precision, recall, f1#, fp, fn
#%%
#precision, recall, f1, fp, fn = evaluate(pred_pos, pos)
precision, recall, f1 = evaluate(pred_pos, pos)
print("Threshold = \t{}\nPrecision = \t{}\nRecall = \t{}\nF1-score = \t{}"
.format(THRESHOLD, precision, recall, f1))
print("{}\t{}\t{}\t{}"
.format(THRESHOLD, precision, recall, f1))
#with open("./Results/ThreshldOpt.csv", "a", newline='') as f:
# writer = csv.writer(f)
# writer.writerow([THRESHOLD, precision, recall, f1])
#%%
#fileHeader = ["Threshold", "Precision", "Recall", "F1-score"]
#with open("./Results/ThreshldOpt.csv", "w") as f:
# writer = csv.writer(f)
# writer.writerow(fileHeader)
#%% find failed/false
# =============================================================================
# import cv2
#
# for img_name in val_set:
# img = cv2.imread('../VOC2007/JPEGImages/' + img_name + '.jpg')
#
# i = 0
# for center in fn:
# box = [center[0] - int(w/2), center[1] - int(w/2), center[0] + int(w/2), center[1] + int(w/2)]
# cropped = img[box[1]:box[1]+w, box[0]:box[0]+w] # [Ymin:Ymax , Xmin:Xmax]
# cv2.imwrite("../VOC2007/Patches/Prediction/FN/fn_{}_{}.jpg".format(img_name, '0'*(4 - len(str(i))) + str(i)),
# cropped, [int(cv2.IMWRITE_JPEG_QUALITY), 100])
# i += 1
#
# j = 0
# for center in fp:
# box = [center[0] - int(w/2), center[1] - int(w/2), center[0] + int(w/2), center[1] + int(w/2)]
# cropped = img[box[1]:box[1]+w, box[0]:box[0]+w] # [Ymin:Ymax , Xmin:Xmax]
# cv2.imwrite("../VOC2007/Patches/Prediction/FP/fn_{}_{}.jpg".format(img_name, '0'*(4 - len(str(j))) + str(j)),
# cropped, [int(cv2.IMWRITE_JPEG_QUALITY), 100])
# j += 1
# =============================================================================