def getFeatureFromTorch(lfw_dir, feature_save_dir, resume=None, gpu=True):
net = model.MobileFacenet()
if gpu:
net = net.to(device)
if resume:
if gpu:
ckpt = torch.load(resume)
else:
ckpt = torch.load(resume, map_location='cpu')
net.load_state_dict(ckpt['net_state_dict'])
net.eval()
nl, nr, flods, flags = parseList(lfw_dir)
lfw_dataset = LFW(nl, nr)
lfw_loader = torch.utils.data.DataLoader(lfw_dataset,
batch_size=32,
shuffle=False,
num_workers=8,
drop_last=False)
featureLs = None
featureRs = None
count = 0
is_calc_time = True
ans = 0
for data in tqdm(lfw_loader):
if gpu:
for i in range(len(data)):
data[i] = data[i].to(device)
count += data[0].size(0) # batch的大小
print('extracing deep features from the face pair {}...'.format(count))
if is_calc_time:
start_time = time.time()
res = [net(d).data.cpu().numpy() for d in data]
end_time = time.time()
is_calc_time = False
# 算出平均每张图片的毫秒数量
count = len(data) * len(data[0]) # 图片总数量
ans = int(((end_time - start_time) * 1000 + 0.5) / count)
else:
res = [net(d).data.cpu().numpy() for d in data]
featureL = np.concatenate((res[0], res[1]), 1)
featureR = np.concatenate((res[2], res[3]), 1)
if featureLs is None:
featureLs = featureL
else:
featureLs = np.concatenate((featureLs, featureL), 0)
if featureRs is None:
featureRs = featureR
else:
featureRs = np.concatenate((featureRs, featureR), 0)
result = {'fl': featureLs, 'fr': featureRs, 'fold': flods, 'flag': flags}
scipy.io.savemat(feature_save_dir, result)
return ans
def getFeatureFromTorch(lfw_dir, feature_save_dir, resume=None, gpu=True):
net = model.ShuffleFaceNet()
if gpu:
net = net.cuda()
if resume:
ckpt = torch.load(resume)
net.load_state_dict(ckpt['net_state_dict'])
net.eval()
nl, nr, flods, flags = parseList(lfw_dir)
lfw_dataset = LFW(nl, nr)
lfw_loader = torch.utils.data.DataLoader(lfw_dataset,
batch_size=32,
shuffle=False,
num_workers=8,
drop_last=False)
featureLs = None
featureRs = None
count = 0
for data in lfw_loader:
if gpu:
for i in range(len(data)):
data[i] = data[i].cuda()
count += data[0].size(0)
print('extracing deep features from the face pair {}...'.format(count))
start_time = time.time()
res = [net(d).data.cpu().numpy() for d in data]
featureL = np.concatenate((res[0], res[1]), 1)
featureR = np.concatenate((res[2], res[3]), 1)
if featureLs is None:
featureLs = featureL
else:
featureLs = np.concatenate((featureLs, featureL), 0)
if featureRs is None:
featureRs = featureR
else:
featureRs = np.concatenate((featureRs, featureR), 0)
# featureLs.append(featureL)
# featureRs.append(featureR)
fps_str = "FPS: %.2f" % (1 / (time.time() - start_time))
print(fps_str)
result = {'fl': featureLs, 'fr': featureRs, 'fold': flods, 'flag': flags}
scipy.io.savemat(feature_save_dir, result)
os.makedirs(save_dir)
logging = init_log(save_dir)
_print = logging.info
# define trainloader and testloader
trainset = CASIA_Face(root=CASIA_DATA_DIR)
trainloader = torch.utils.data.DataLoader(trainset,
batch_size=BATCH_SIZE,
shuffle=True,
num_workers=8,
drop_last=False)
# nl: left_image_path
# nr: right_image_path
nl, nr, folds, flags = parseList(root=LFW_DATA_DIR)
testdataset = LFW(nl, nr)
testloader = torch.utils.data.DataLoader(testdataset,
batch_size=32,
shuffle=False,
num_workers=8,
drop_last=False)
# define model
net = model.MobileFacenet()
ArcMargin = model.ArcMarginProduct(128, trainset.class_nums)
if RESUME:
ckpt = torch.load(RESUME)
net.load_state_dict(ckpt['net_state_dict'])
start_epoch = ckpt['epoch'] + 1
def main():
# 加载训练好的模型
net = model.MobileFacenet()
ckpt = torch.load("model/best/068.ckpt", map_location='cpu')
net.load_state_dict(ckpt['net_state_dict'])
net.eval()
# 加载数据集
nl, nr, flods, flags = parseList(LFW_DATA_DIR) # 得到左边人和右边人
lfw_dataset = LFW(nl, nr)
lfw_loader = torch.utils.data.DataLoader(lfw_dataset,
batch_size=32,
shuffle=False,
num_workers=8,
drop_last=False)
featureLs = None
featureRs = None
count = 0 # 图片总数
# 这一步是在网络算出每个人的人脸特征，得到了左边的人脸特征和右边的特征，做余弦相似度，就能判断是不是同一个人
for i, data in tqdm(enumerate(lfw_loader)):
# data (4， 32, 3, 112, 96)
count += data[0].size(0) # batch的大小
# print('extracing deep features from the face pair {}...'.format(count))
res = [net(d).data.cpu().numpy() for d in data] # d 表示一个图像
featureL = np.concatenate((res[0], res[1]), 1) # 纵向拼接，也就是接在列后面
featureR = np.concatenate((res[2], res[3]), 1)
if featureLs is None:
featureLs = featureL
else:
featureLs = np.concatenate((featureLs, featureL),
0) # 横向拼接，也就是接在行后面
if featureRs is None:
featureRs = featureR
else:
featureRs = np.concatenate((featureRs, featureR), 0)
# print(featureRs.shape) # （6000， 256）这里256的原因是在加载数据集时将逆转宽度的向量也加入进来了，类似与得到了一张左右对称的图
result = {'fl': featureLs, 'fr': featureRs, 'fold': flods, 'flag': flags}
scipy.io.savemat("result.mat", result)
ACCs = np.zeros(10)
result = scipy.io.loadmat("result.mat")
for i in range(10):
fold = result['fold']
flags = result['flag']
featureLs = result['fl']
featureRs = result['fr']
valFold = fold != i
testFold = fold == i
flags = np.squeeze(flags)
mu = np.mean(
np.concatenate(
(featureLs[valFold[0], :], featureRs[valFold[0], :]), 0),
0) # 获得人脸特征的均值
mu = np.expand_dims(mu, 0)
featureLs = featureLs - mu # 减去均值
featureRs = featureRs - mu
featureLs = featureLs / np.expand_dims(
np.sqrt(np.sum(np.power(featureLs, 2), 1)), 1) # 除以根号下的平方
featureRs = featureRs / np.expand_dims(
np.sqrt(np.sum(np.power(featureRs, 2), 1)), 1)
# 其实上面几步就是归一化，映射到-1到1之间
scores = np.sum(np.multiply(featureLs, featureRs), 1) # 求向量内积
threshold = getThreshold(scores[valFold[0]], flags[valFold[0]],
10000) # 从验证集中获得精度最高的阈值
ACCs[i] = getAccuracy(scores[testFold[0]], flags[testFold[0]],
threshold) # 根据阈值，算在测试机中的准确率
return ACCs