-
Notifications
You must be signed in to change notification settings - Fork 25
/
evaluation_old.py
212 lines (162 loc) · 6.7 KB
/
evaluation_old.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
import numpy as np
import os
import torch
import torch.nn as nn
from scipy.spatial.distance import cdist
from torch.autograd import Variable
class Evaluation():
def __init__(self, df_test, df_query, dataloader_test, dataloader_query, cuda):
self.test_labels, self.test_camera_labels, self.test_image_paths, self.test_image_names = get_info(df_test)
self.query_labels, self.query_camera_labels, self.query_image_paths, self.query_image_names = get_info(df_query)
self.Index = make_index(self.test_image_names, self.query_image_names)
self.dataloader_test = dataloader_test
self.dataloader_query = dataloader_query
self.cuda = cuda
def ranks_map(self, model, nr, remove_fc=False):
if remove_fc:
model = nn.Sequential(*list(model.children())[:-1])
gescr_gallery = get_descr(self.dataloader_test, model, self.cuda)
gescr_query = get_descr(self.dataloader_query, model, self.cuda)
ranks, img_names_sorted, places = ranking('cosine', gescr_query, self.query_image_names, gescr_gallery, self.test_image_names, nr, self.Index)
ranks_w = ranks / len(gescr_query) * 1.0
mAP_ = mAP(gescr_query, self.query_image_names, gescr_gallery, self.test_image_names, nr, self.Index)
return ranks_w, mAP_
def get_descr(dataloder, model, use_gpu):
result = []
for data in dataloder:
if use_gpu:
inputs = Variable(data.cuda())
else:
inputs = Variable(data)
out = model(inputs)
result.extend(out.data.squeeze().cpu().numpy())
return np.array(result)
def get_info(df):
labels = np.array(df['label'])
camera_labels = np.array(df['camera'])
image_paths = np.array(df['path'])
image_names = np.array(df['name'])
return labels, camera_labels, image_paths, image_names
def make_index(test_image_names, query_image_names):
Index = dict()
Index['junk'] = set()
Index['distractor'] = set()
for name in test_image_names:
if ifJunk(name):
Index['junk'].add(name)
elif ifDistractor(name):
Index['distractor'].add(name)
for query in query_image_names:
Index[query] = dict()
Index[query]['pos'] = set()
Index[query]['junk'] = set()
person, camera = parse_market_1501_name(query)
for name in test_image_names:
if ifJunk(name) or ifDistractor(name):
continue
person_, camera_ = parse_market_1501_name(name)
if person == person_ and camera != camera_ :
Index[query]['pos'].add(name)
elif person == person_ and camera == camera :
Index[query]['junk'].add(name)
return Index
def parse_market_1501_name(full_name):
name_ar = full_name.split('/')
name = name_ar[len(name_ar)-1]
person = int(name.split('_')[0])
camera = int(name.split('_')[1].split('s')[0].split('c')[1])
return person, camera
def parseMarket1501(path):
person_label = []
camera_label = []
image_path = []
image_name = []
for file in sorted(os.listdir(path)):
if file.endswith(".jpg"):
person, camera = parse_market_1501_name(file)
person_label.append(person)
camera_label.append(camera)
image_path.append(os.path.join(path, file))
image_name.append(file)
return person_label, camera_label, image_path, image_name
def ifJunk(filename):
if filename.startswith("-1"):
return True
else:
return False
def ifDistractor(filename):
if filename.startswith("0000"):
return True
else:
return False
def getPlace(query, sorted_gallery_filenames, Index):
place = 0
for i in range(len(sorted_gallery_filenames)):
if sorted_gallery_filenames[i] in Index['junk'] or sorted_gallery_filenames[i] in Index[query]['junk']:
continue
elif sorted_gallery_filenames[i] in Index['distractor']:
place +=1
elif sorted_gallery_filenames[i] in Index[query]['pos']:
# print "PLACE " , sorted_gallery_filenames[i]
return place
else :
place +=1
return place
def ranking(metric, gescrs_query, query_image_names, gescrs_gallery, test_image_names, maxrank, Index):
ranks = np.zeros(maxrank + 1)
places = dict()
all_dist = cdist(gescrs_query, gescrs_gallery, metric)
np_test_image_names = np.array(test_image_names)
img_names_sorted = dict()
all_gallery_names_sorted = np_test_image_names[np.argsort(all_dist).astype(np.uint32)]
for qind in range(len(gescrs_query)):
dist = all_dist[qind]
gallery_names_sorted = all_gallery_names_sorted[qind]
place=getPlace(query_image_names[qind], gallery_names_sorted, Index)
img_names_sorted[qind] = all_gallery_names_sorted[qind]
places[qind] = place
ranks[place+1:maxrank+1] += 1
return ranks, img_names_sorted,places
def cos_dist(x, y):
xy = np.dot(x,y);
xx = np.dot(x,x);
yy = np.dot(y,y);
return -xy*1.0/np.sqrt(xx*yy)
def getDistances(gescr_query, gescrs_gallery):
dist = list()
for i in range(len(gescrs_gallery)):
dist.append(cos_dist(gescr_query, gescrs_gallery[i]))
return dist
def getAveragePrecision(query, sorted_gallery_filenames, Index):
ap = 0
tp = 0
k = 0
for i in range(len(sorted_gallery_filenames)):
if sorted_gallery_filenames[i] in Index['junk'] or sorted_gallery_filenames[i] in Index[query]['junk']:
continue
elif sorted_gallery_filenames[i] in Index['distractor']:
k+=1
deltaR = 0
elif sorted_gallery_filenames[i] in Index[query]['pos']:
tp+=1
k+=1
deltaR = 1.0/len(Index[query]['pos'])
else :
k +=1
deltaR = 0
precision = tp*1.0/k * deltaR
ap += precision
if tp == len(Index[query]['pos']):
return ap
return ap
def mAP(gescrs_query, query_image_names, gescrs_gallery, test_image_names, maxrank, Index):
ranks = np.zeros(maxrank+1)
places = dict()
ap_sum = 0
for qind in range(len(gescrs_query)):
dist = getDistances(gescrs_query[qind], gescrs_gallery)
dist_zip = sorted(zip(dist,test_image_names))
gallery_names_sorted = [x for (y,x) in dist_zip]
ap=getAveragePrecision(query_image_names[qind], gallery_names_sorted, Index)
ap_sum += ap
return ap_sum * 1.0 /len(gescrs_query)