-
Notifications
You must be signed in to change notification settings - Fork 19
/
Copy pathdata.py
180 lines (162 loc) · 5.69 KB
/
data.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
"""
create by Qiang Zhang
function: provide datasets
"""
import os
import numpy as np
import random
import torch.utils.data as Data
from utils import readpkl,select_p,select_p_n
from config import data_root
class RawTrain_core(Data.Dataset):
def __init__(self,thresh):
self.traindata = Traindata(thresh)
def __getitem__(self, id0):
id = self.traindata.keys[id0]
items = self.traindata.data[id]
data = []
for item in items:
det = self.traindata.det[item[0]]
feat = self.traindata.feat[item[0]]
roi = item[1]
feat1,feat3 = select_p(det,feat,roi)
# data.append(feat3)
data.append(np.asarray([feat1,feat1,feat1]))
return id,np.asarray(data)
def __len__(self):
return self.traindata.__len__()
class RawTrain(Data.Dataset):
def __init__(self,thresh):
self.data_id = RawTrain_core(thresh)
self.core()
def core(self):
data = []
for id,item in enumerate(self.data_id):
for person in item[1]:
data.append([item[0],person])
self.data = data
def __getitem__(self, id):
return self.data[id]
def __len__(self):
return len(self.data)
@classmethod
def getLoader(cls,thresh=0.5,batch_size=2):
dataset = cls(thresh)
return Data.DataLoader(
dataset = dataset,
batch_size=batch_size,
shuffle=True,
num_workers=2
)
class Traindata(Data.Dataset):
"""
feat1:only the probe person feature:1 x 256
feat3:the probe person and his two partners:3 x 256
label:true or false:1 x 256
gallery_all:all of the gallery features:100 x n x 256
"""
def __init__(self,thresh,neibor=4):
self.neibor = neibor
self.root = os.path.join(data_root,'predata/traindata')
self.traindata = readpkl(self.root + "/psdb_train_gt_roidb.pkl")
traindetections = readpkl(self.root + "/gallery_detections.pkl")
trainfeatures = readpkl(self.root + "/gallery_features.pkl")
self.filter(traindetections,trainfeatures['feat'],thresh)
self.data = self.getIndex()
self.keys = list(self.data.keys())
self.sample = list(range(len(self.det)))
def filter(self,traindetections,trainfeatures,thresh=0.0):
self.det = []
self.feat = []
for item,item2 in zip(traindetections,trainfeatures):
ids = item[:,4]>thresh
item1 = item[ids]
item2 = item2[ids,:,0,0]
self.det.append(item1)
self.feat.append(item2)
def getIndex(self):
personSet = {}
for i in range(11204):
ids = self.traindata[i + 1]['gt_pids']
detection = self.det[i + 1]
if len(detection)==0:
continue
for j,person in enumerate(ids):
if person==-1:
continue
try:
tmp = [i+1,detection[j]]
except:
continue
try:
personSet[person].append(tmp)
except:
personSet[person] = [tmp]
return personSet
def getGallery(self,probe):
slice = random.sample(self.sample,100)
label = [0]*100
gallery = []
for i in range(100):
tmp = self.feat[slice[i]]
gallery.append(tmp)
if len(probe)==1:
return label,gallery
for i,item in enumerate(probe[1:]):
if item[0] in slice:
id = slice.index(item[0])
tmp = random.sample(self.sample,1)[0]
gallery[id] = self.feat[tmp]
gallery[i] = self.feat[item[0]]
label[i] = 1
if not sum(label) == sum(label[:sum(label)]):
print("error")
return label,gallery
def getGalleryRandom(self,probe, randIdx):
slice = random.sample(self.sample,100)
label = [0]*100
gallery = []
idx = list(range(len(probe)))
idx.remove(randIdx)
for i in range(100):
tmp = self.feat[slice[i]]
gallery.append(tmp)
if len(probe)==1:
return label,gallery
for i, ix in enumerate(idx):
item = probe[ix]
if item[0] in slice:
id = slice.index(item[0])
tmp = random.sample(self.sample,1)[0]
gallery[id] = self.feat[tmp]
gallery[i] = self.feat[item[0]]
label[i] = 1
if not sum(label) == sum(label[:sum(label)]):
print("error")
return label,gallery
def __getitem__(self, id):
id = self.keys[id]
randIdx = random.randint(0, len(self.data[id])-1)
item = self.data[id][randIdx]
det = self.det[item[0]]
feat = self.feat[item[0]]
roi = item[1]
feat1, featn = select_p_n(det, feat, roi, num=self.neibor)
label, gallery_all = self.getGalleryRandom(self.data[id], randIdx)
return feat1, featn, np.asarray(label), np.asarray(gallery_all)
'''
id = self.keys[id]
item = self.data[id][0]
det = self.det[item[0]]
feat = self.feat[item[0]]
roi = item[1]
feat1,featn = select_p_n(det,feat,roi,num=self.neibor)
label,gallery_all = self.getGallery(self.data[id])
return feat1,featn,np.asarray(label),np.asarray(gallery_all)
'''
def __len__(self):
return len(self.data)
def getTestdata(neibor=4):
testFile = os.path.join(data_root,'testdata/testdata_featn_{}.npy'.format(neibor))
testData = np.load(testFile,encoding='latin1',allow_pickle = True)
return testData