/
VLAD_Performance.py
65 lines (62 loc) · 2.65 KB
/
VLAD_Performance.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
from vlad_descriptor import *
import numpy as np
import pickle
from utils import vlad_group_representation, partition_data, hashing
from scipy.stats import zscore
if __name__ == '__main__':
with open('dataset_LFW_VGG3.pkl', 'rb') as handle:
dataset = pickle.load(handle)
np.random.seed(10)
n_Clus = 4
dim = len(dataset['data_x'][0]) * n_Clus
l = int(dim)
Ptp01, Ptp05 = [], []
S_x = int(l*0.7) # 610
group_member = 40 # 2,5,10,16,20,40
W = np.random.random([dim, l])
U, S, V = np.linalg.svd(W)
W = U[:, :l]
param = {'encode': 'soft', 'normalize': 3, 'n_Clus': n_Clus}
# Assign data to groups
data = np.array(dataset['data_x'])
data = zscore(np.array(dataset['data_x']), axis=0) # N*d
groups = partition_data(data, group_member, partitioning='random')
# Compute group representations
group_vec, VLAD_Codebook = vlad_group_representation(data, groups, param)
group_vec = np.array(group_vec).T
group_vec = hashing(group_vec, W, S_x)
# The embedding for H0 queries
n_q0 = len(dataset['H0_id'])
H0_data = zscore(np.array(dataset['H0_x']).T, axis=1) # LFW
# H0_data = zscore(np.array(dataset['H0_x']), axis=0) # CFP
H0_data = [VLADEncoding(np.expand_dims(H0_data[i, :], axis=0), VLAD_Codebook,
encode=param['encode'], normalize=param['normalize']) for i in range(n_q0)]
Q0 = np.array(H0_data).T
Q0 = hashing(np.array(H0_data).T, W, S_x)
H0_claimed_id = np.random.randint(0, len(groups['ind']), size=n_q0).astype(np.int)
D00 = np.linalg.norm(Q0 - group_vec[:, H0_claimed_id], axis=0)
# The embedding for H1 queries
n_q1 = len(dataset['H1_id'])
H1_group_id = np.zeros(n_q1)
H1_data = zscore(np.array(dataset['H1_x']), axis=0)
H1_data = [VLADEncoding(np.expand_dims(H1_data[i, :], axis=0), VLAD_Codebook,
encode=param['encode'], normalize=param['normalize']) for i in range(n_q1)]
Q1 = np.array(H1_data).T
Q1 = hashing(np.array(H1_data).T, W, S_x)
# Determine the group identity of H1 queries
for i in range(len(groups['ind'])):
group_id = [dataset['data_id'][x] for x in groups['ind'][i]]
a = [n for n, x in enumerate(dataset['H1_id']) for y in group_id if x == y]
for x in a:
H1_group_id[x] = i
D11 = np.linalg.norm(Q1 - group_vec[:, H1_group_id.astype(np.int)], axis=0)
D0 = np.sort(D00)
D1 = np.sort(D11)
Pfp = 0.01
tau = D0[int(Pfp * n_q0)]
Ptp01 = np.count_nonzero(D1 <= tau) / n_q1
Pfp = 0.05
tau = D0[int(Pfp * n_q0)]
Ptp05 = np.count_nonzero(D1 <= tau) / n_q1
print('Ptp01:', Ptp01, 'Ptp05', Ptp05)
a = 2