-
Notifications
You must be signed in to change notification settings - Fork 1
/
evaluate.py
176 lines (160 loc) · 6.58 KB
/
evaluate.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
from sklearn.preprocessing import label_binarize
from sklearn.metrics import roc_curve, auc, confusion_matrix
import numpy as np
import tensorflow as tf
import os, sys
from utils import process_i as process
config = tf.ConfigProto()
config.gpu_options.allow_growth = True
nb_nodes = 20
nhood = 1
ft_size = 3
time_step = 6
batch_size = 256
lr = 0.005 # learning rate
tf.app.flags.DEFINE_string('dataset', 'KS20', "Dataset: IAS, KS20 or KGBD")
tf.app.flags.DEFINE_string('length', '6', "4, 6, 8 or 10")
tf.app.flags.DEFINE_string('split', '', "for IAS-Lab testing splits (A or B)")
tf.app.flags.DEFINE_string('gpu', '0', "GPU number")
tf.app.flags.DEFINE_string('model_dir', 'best', "model directory") # 'best' will test the best model in current directory
FLAGS = tf.app.flags.FLAGS
# check parameters
if FLAGS.dataset not in ['IAS', 'KGBD', 'KS20']:
raise Exception('Dataset must be IAS, KGBD, or KS20.')
if not FLAGS.gpu.isdigit() or int(FLAGS.gpu) < 0:
raise Exception('GPU number must be a positive integer.')
if FLAGS.length not in ['4', '6', '8', '10']:
raise Exception('Length number must be 4, 6, 8 or 10.')
if FLAGS.split not in ['', 'A', 'B']:
raise Exception('Datset split must be "A" (for IAS-A), "B" (for IAS-B), "" (for other datasets).')
os.environ["CUDA_VISIBLE_DEVICES"] = FLAGS.gpu
dataset = FLAGS.dataset
time_step = int(FLAGS.length)
split = FLAGS.split
model_dir = FLAGS.model_dir
def evaluate_reid(model_dir, dataset):
if dataset == 'IAS':
classes = list(range(11))
elif dataset == 'KS20':
classes = list(range(20))
elif dataset == 'KGBD':
classes = list(range(164))
checkpoint = model_dir + ".ckpt"
print('Evaluating the model saved in ' + model_dir)
loaded_graph = tf.get_default_graph()
with tf.Session(graph=loaded_graph, config=config) as sess:
loader = tf.train.import_meta_graph(checkpoint + '.meta')
loader.restore(sess, checkpoint)
# loader = tf.train.import_meta_graph(checkpt_file + '.meta')
# loader.restore(sess, checkpt_file)
J_in = loaded_graph.get_tensor_by_name("Input/Placeholder_1:0")
P_in = loaded_graph.get_tensor_by_name("Input/Placeholder_2:0")
B_in = loaded_graph.get_tensor_by_name("Input/Placeholder_3:0")
I_in = loaded_graph.get_tensor_by_name("Input/Placeholder_4:0")
J_bias_in = loaded_graph.get_tensor_by_name("Input/Placeholder_5:0")
P_bias_in = loaded_graph.get_tensor_by_name("Input/Placeholder_6:0")
B_bias_in = loaded_graph.get_tensor_by_name("Input/Placeholder_7:0")
I_bias_in = loaded_graph.get_tensor_by_name("Input/Placeholder_8:0")
lbl_in = loaded_graph.get_tensor_by_name("Input/Placeholder:0")
is_train = loaded_graph.get_tensor_by_name("Input/Placeholder_11:0")
attn_drop = loaded_graph.get_tensor_by_name("Input/Placeholder_9:0")
ffd_drop = loaded_graph.get_tensor_by_name("Input/Placeholder_10:0")
aver_pre = loaded_graph.get_tensor_by_name('Recognition/Recognition/add_30:0')
accuracy = loaded_graph.get_tensor_by_name('Recognition/Recognition/Mean_4:0')
loss = loaded_graph.get_tensor_by_name('Recognition/Recognition/Mean_5:0')
rank_acc = {}
en_to_pred = loaded_graph.get_tensor_by_name("Recognition/Recognition/StopGradient:0")
X_train_J, X_train_P, X_train_B, X_train_I, y_train, X_test_J, X_test_P, X_test_B, X_test_I, y_test, \
adj_J, biases_J, adj_P, biases_P, adj_B, biases_B, adj_I, biases_I, nb_classes = \
process.gen_train_data(dataset=dataset, split=split, time_step=time_step,
nb_nodes=nb_nodes, nhood=nhood, global_att=False, batch_size=batch_size, view='',
reverse='0')
# print(batch_size)
X_train = X_train_J
X_test = X_test_J
vl_step = 0
vl_size = X_test.shape[0]
logits_all = []
labels_all = []
vl_step = 0
vl_loss = 0.0
vl_acc = 0.0
while vl_step * batch_size < vl_size:
if (vl_step + 1) * batch_size > vl_size:
break
X_input_J = X_test_J[vl_step * batch_size:(vl_step + 1) * batch_size]
X_input_J = X_input_J.reshape([-1, nb_nodes, 3])
X_input_P = X_test_P[vl_step * batch_size:(vl_step + 1) * batch_size]
X_input_P = X_input_P.reshape([-1, 10, 3])
X_input_B = X_test_B[vl_step * batch_size:(vl_step + 1) * batch_size]
X_input_B = X_input_B.reshape([-1, 5, 3])
X_input_I = X_test_I[vl_step * batch_size:(vl_step + 1) * batch_size]
X_input_I = X_input_I.reshape([-1, I_nodes, 3])
y_input = y_test[vl_step * batch_size:(vl_step + 1) * batch_size]
loss_value_vl, acc_vl, pred = sess.run([loss, accuracy, aver_pre],
feed_dict={
J_in: X_input_J,
P_in: X_input_P,
B_in: X_input_B,
I_in: X_input_I,
J_bias_in: biases_J,
P_bias_in: biases_P,
B_bias_in: biases_B,
I_bias_in: biases_I,
lbl_in: y_test[vl_step * batch_size:(vl_step + 1) * batch_size],
is_train: False,
attn_drop: 0.0, ffd_drop: 0.0})
for i in range(y_input.shape[0]):
for K in range(1, len(classes) + 1):
if K not in rank_acc.keys():
rank_acc[K] = 0
t = np.argpartition(pred[i], -K)[-K:]
if np.argmax(y_input[i]) in t:
rank_acc[K] += 1
logits_all.extend(pred.tolist())
labels_all.extend(y_input.tolist())
vl_loss += loss_value_vl
vl_acc += acc_vl
vl_step += 1
for K in rank_acc.keys():
rank_acc[K] /= (vl_step * batch_size)
rank_acc[K] = round(rank_acc[K], 4)
val_nAUC = process.cal_nAUC(scores=np.array(logits_all), labels=np.array(labels_all))
from sklearn.metrics import roc_curve, auc, confusion_matrix
y_true = np.argmax(np.array(labels_all), axis=-1)
y_pred = np.argmax(np.array(logits_all), axis=-1)
print('\n### Re-ID Confusion Matrix: ')
print(confusion_matrix(y_true, y_pred))
print('### Rank-N Accuracy: ')
print(rank_acc)
print('### Test loss:', round(vl_loss / vl_step, 4), '; Test accuracy:', round(vl_acc / vl_step, 4),
'; Test nAUC:', round(val_nAUC, 4))
exit()
if dataset == 'KS20':
nb_nodes = 25
I_nodes = 49
batch_size = 64
elif dataset == 'IAS' or dataset == 'KGBD':
nb_nodes = 20
I_nodes = 39
elif dataset == 'CASIA_B':
nb_nodes = 14
I_nodes = 27
batch_size = 128
if split == 'A':
batch_size = 128
elif split == 'B':
batch_size = 64
if dataset == 'KGBD':
batch_size = 256
if model_dir == 'best':
if dataset == 'IAS' and split == 'A':
batch_size = 64
model_dir = 'RN/IAS-A_59.4_86.7_formal'
elif dataset == 'IAS' and split == 'B':
model_dir = 'RN/IAS-B_69.8_90.4_formal'
elif dataset == 'KS20':
model_dir = 'RN/KS20_87.5_95.8_formal'
elif dataset == 'KGBD':
model_dir = 'RN/KGBD_99.5_99.6_formal'
evaluate_reid(model_dir, dataset)