-
Notifications
You must be signed in to change notification settings - Fork 1
/
lfw_arc_test.py
203 lines (158 loc) · 7.72 KB
/
lfw_arc_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
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
import argparse
import numpy as np
from PIL import Image
import torch
import torch.utils.data
import torch.optim
import torchvision.transforms as transforms
import torch.backends.cudnn as cudnn
cudnn.benchmark = True
from lib.core.config import config
from lib.core.config import update_config
from lib.core.lfw_eval import eval as lfw_eval
from lib.core.lfw_eval import eval_roc
from lib.core.lfw_eval import extractDeepFeatureSingle, cosine_similarity
from lib.datasets.dataset import LFW_Image
# from model.CBAM import CBAMResNet
from arcface.face_model import MobileFaceNet
from utils import Visualizer
# setup random seed
torch.manual_seed(0)
np.random.seed(0)
def parse_args():
parser = argparse.ArgumentParser(description='Pytorch End2End Occluded Face')
parser.add_argument('--cfg', help='experiment configure file name', required=True, type=str)
args, rest = parser.parse_known_args()
update_config(args.cfg)
parser.add_argument('--gpus', help='gpus', type=str)
parser.add_argument('--workers', help='num of dataloader workers', type=int)
parser.add_argument('--binary_thres', help='thres for binary mask', type=float)
parser.add_argument('--soft_binary', help='whether use soft binary mask', type=int)
parser.add_argument('--batch_size', help='batch size', type=int)
parser.add_argument('--pretrained', help='whether use pretrained model', type=str)
parser.add_argument('--debug', help='whether debug', default=0, type=int)
parser.add_argument('--model', help=' model name', type=str)
parser.add_argument('--factor', help='factor of mask', type=float)
parser.add_argument('--ratio', help='ratio of masked img for training', default=4, type=int)
args = parser.parse_args()
return args
def plot_roc(occ_paths, visualizer, append=False):
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
# --------------------------------------model----------------------------------------
test_transform = transforms.Compose([
transforms.ToTensor(), # range [0, 255] -> [0.0,1.0]
transforms.Normalize(mean=(0.5, 0.5, 0.5), std=(0.5, 0.5, 0.5)) # range [0.0, 1.0] -> [-1.0,1.0]
])
test_loaders = []
# occ_paths = ['data_occ/lfws1/random_block_occ/', 'data_occ/lfws1/random_block_ca_fill/',
# 'data_occ/lfws1/random_block_pic_fill/', 'data_occ/lfws1/random_block_ca_id_fill/']
# occ_paths = ['data_occ/lfws1/random_part_occ/', 'data_occ/lfws1/random_part_ca_fill/',
# 'data_occ/lfws1/random_part_pic_fill/', 'data_occ/lfws1/random_part_ca_id_fill/']
for p in occ_paths:
config.DATASET.LFW_PATH = 'data_occ/lfws/lfw_align_112/'
config.DATASET.LFW_OCC_PATH = p
config.DATASET.LFW_PAIRS = 'data_occ/lfws/pairs.txt'
test_loader = torch.utils.data.DataLoader(
LFW_Image(config, test_transform),
batch_size=config.TEST.BATCH_SIZE,
shuffle=config.TEST.SHUFFLE,
num_workers=config.TEST.WORKERS,
pin_memory=True)
test_loaders.append(test_loader)
model_list = ['Arc. ']
for model_name in model_list:
recognizer = MobileFaceNet(512).to(device) # embeding size is 512 (feature vector)
recognizer.load_state_dict(torch.load('arcface/MobileFace_Net', map_location=lambda storage, loc: storage))
recognizer.to(device)
recognizer.eval()
# visualizer = Visualizer()
eval_roc(recognizer, model_name, config, test_loaders, 'temp', 0, visualizer=visualizer, append=append)
def plot():
occ_paths = ['data/LFWs/20220712/random_block_occ/', 'data/LFWs/20220712/random_block_ca_fill/',
'data/LFWs/20220712/random_block_pic_fill/', 'data/LFWs/20220712/random_block_ca_id_fill/']
visualizer = Visualizer()
plot_roc(occ_paths, visualizer, True)
# from_roc(paths, visualizer, True)
def main():
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
# --------------------------------------model----------------------------------------
test_transform = transforms.Compose([
transforms.ToTensor(), # range [0, 255] -> [0.0,1.0]
transforms.Normalize(mean=(0.5, 0.5, 0.5), std=(0.5, 0.5, 0.5)) # range [0.0, 1.0] -> [-1.0,1.0]
])
# config.DATASET.LFW_PATH = 'data_occ/CFP-FP/cfp-crop/Data/Images/'
# config.DATASET.LFW_OCC_PATH = 'data_occ/CFP-FP/ca_id_fill_mask/Data/Images/'
# config.DATASET.LFW_PAIRS = 'data_occ/CFP-FP/cfp-crop/Protocol/Pair_list_F.txt'
# config.DATASET.LFW_CLASS = 6000
#
# test_loader = torch.utils.data.DataLoader(
# CFP_Image(config, test_transform),
# batch_size=config.TEST.BATCH_SIZE,
# shuffle=config.TEST.SHUFFLE,
# num_workers=config.TEST.WORKERS,
# pin_memory=True)
config.DATASET.LFW_PATH = 'data/LFWs/lfw_112/'
config.DATASET.LFW_OCC_PATH = 'data/LFWs/20220712/random_part_ca_id_fill/'
config.DATASET.LFW_PAIRS = 'data/LFWs/pairs.txt'
test_loader = torch.utils.data.DataLoader(
LFW_Image(config, test_transform),
batch_size=config.TEST.BATCH_SIZE,
shuffle=config.TEST.SHUFFLE,
num_workers=config.TEST.WORKERS,
pin_memory=True)
model_root = 'pretrained/'
# model_list = ['model_p5_w1_9938_9470_6503.pth.tar',
# 'model_p4_baseline_9938_8205_3610.pth.tar']
model_list = ['arcface']
# model_list = [
# 'model_best_p5_w0.pth.tar',
# 'model_best_p5_w1.pth.tar',
# 'model_best_p5_occ.pth.tar'
# ]
for model_name in model_list:
# pattern = int(model_name[model_name.find('p')+1])
# num_mask = len(utils.get_grids(*config.NETWORK.IMAGE_SIZE, pattern))
# model = LResNet50E_IR_FPN(num_mask=num_mask)
# model = torch.nn.DataParallel(model, device_ids=gpus).cuda()
# recognizer = CBAMResNet(50, feature_dim=512, mode='ir') # resnet18
# recognizer.load_state_dict(torch.load('./arcface/model_ir_se50.pth', map_location=device))
recognizer = MobileFaceNet(512).to(device) # embeding size is 512 (feature vector)
recognizer.load_state_dict(torch.load('arcface/MobileFace_Net', map_location=lambda storage, loc: storage))
# print('MobileFaceNet face detection model generated')
# detect_model.eval()
recognizer.to(device)
recognizer.eval()
lfw_eval(recognizer, model_name, config, test_loader, 'temp', 0)
def get_distance(img1_path, img2_path):
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
recognizer = CBAMResNet(50, feature_dim=512, mode='ir') # resnet18
# recognizer = DataParallel(recognizer)
recognizer.load_state_dict(torch.load('./arcface/Iter_64000_net.pth', map_location=device))
recognizer.to(device)
recognizer.eval()
with open(img1_path, 'rb') as f:
img1 = Image.open(f).convert('RGB')
with open(img2_path, 'rb') as f:
img2 = Image.open(f).convert('RGB')
# print(img1.size)
transform = transforms.Compose([
transforms.ToTensor(), # range [0, 255] -> [0.0,1.0]
transforms.Normalize(mean=(0.5, 0.5, 0.5), std=(0.5, 0.5, 0.5)) # range [0.0, 1.0] -> [-1.0,1.0]
])
img1 = transform(img1).to('cuda').unsqueeze(0)
img2 = transform(img2).to('cuda').unsqueeze(0)
# print(img1.size())
f1 = extractDeepFeatureSingle(img1, recognizer)
f2 = extractDeepFeatureSingle(img2, recognizer)
# print(f2.size())
# distance = cosine_similarity(f1.detach(), f2.detach())
f1 = f1.detach().numpy()
f2 = f2.detach().numpy()
# print(f1.shape, f2.shape)
A = np.sum(f1 * f2)
B = np.linalg.norm(f1, axis=0) * np.linalg.norm(f2, axis=0) + 1e-5
distance = A / B
return 1-distance
if __name__ == '__main__':
# plot()
main()