def _load_model(self, model_path):
# define model
model_name = os.path.basename(model_path)
h_input, w_input, model_type, _ = parse_model_name(model_name)
self.kernel_size = get_kernel(
h_input,
w_input,
)
self.model = MODEL_MAPPING[model_type](
conv6_kernel=self.kernel_size).to(self.device)
# load model weight
state_dict = torch.load(model_path, map_location=self.device)
keys = iter(state_dict)
first_layer_name = keys.__next__()
if first_layer_name.find('module.') >= 0:
from collections import OrderedDict
new_state_dict = OrderedDict()
for key, value in state_dict.items():
name_key = key[7:]
new_state_dict[name_key] = value
self.model.load_state_dict(new_state_dict)
else:
self.model.load_state_dict(state_dict)
return None
def update_config(args, conf):
conf.devices = args.devices
conf.val_size = args.val_size
conf.patch_info = args.patch_info
conf.checkpoint = args.checkpoint
conf.epochs = args.epochs
conf.batch_size = args.batch_size
conf.model_type = args.model_type
w_input, h_input = get_width_height(args.patch_info)
conf.input_size = [h_input, w_input]
conf.kernel_size = get_kernel(h_input, w_input)
conf.device = "cuda:{}".format(
conf.devices[0]) if torch.cuda.is_available() else "cpu"
# resize fourier image size
conf.ft_height = 2 * conf.kernel_size[0]
conf.ft_width = 2 * conf.kernel_size[1]
current_time = datetime.now().strftime('%b%d_%H-%M-%S')
job_name = 'Anti_Spoofing_{}'.format(args.patch_info)
log_path = '{}/{}/{} '.format(conf.log_path, job_name, current_time)
snapshot_dir = '{}/{}'.format(conf.snapshot_dir_path, job_name)
make_if_not_exist(snapshot_dir)
make_if_not_exist(log_path)
conf.model_path = snapshot_dir
conf.log_path = log_path
conf.job_name = job_name
return conf
def update_config(args, conf):
conf.devices = args.devices
conf.patch_info = args.patch_info
w_input, h_input = get_width_height(args.patch_info)
conf.kernel_size = get_kernel(h_input, w_input)
conf.device = "cuda:{}".format(
conf.devices[0]) if torch.cuda.is_available() else "cpu"
# resize fourier image size
# conf.ft_height = 2*conf.kernel_size[0]
# conf.ft_width = 2*conf.kernel_size[1]
# eval_snapshot_dir = '{}/{}'.format(conf.eval_snapshot_dir_path, eval_job_name)
make_if_not_exist(conf.model_path)
make_if_not_exist(conf.log_path)
return conf
def __init__(self, **kwargs):
super(OnnxModel, self).__init__()
kernel_size = get_kernel(80, 60)
# self.model = MiniFASNetV2(conv6_kernel = kernel_size)
self.model = RepVggFTNet(32, [48, 64, 128], [4, 10, 8],
num_out=2,
num_classes=3)
try:
# self.model.load("work_dirs/RepVGGNet/epoch_77.pth")
clean_state_dict = OrderedDict()
state_dict = torch.load("work_dirs/RepVGGNet/epoch_77.pth")
for name, value in state_dict.items():
if name.startswith("module."):
clean_state_dict[name.lstrip("module.")] = value
self.model.load_state_dict(clean_state_dict)
load(self.model)
except:
print(
"There is no model loaded!! Initial parameters will be saved")
def update_config(args, conf):
conf.devices = args.devices
conf.patch_info = args.patch_info
w_input, h_input = get_width_height(args.patch_info)
conf.input_size = [h_input, w_input]
conf.kernel_size = get_kernel(w_input, h_input)
conf.device = "cuda:{}".format(
conf.devices[0]) if torch.cuda.is_available() else "cpu"
current_time = datetime.now().strftime('%b%d_%H-%M-%S')
job_name = 'Anti_Spoofing_{}'.format(args.patch_info)
log_path = '{}/{}/{} '.format(conf.log_path, job_name, current_time)
snapshot_dir = '{}/{}'.format(conf.snapshot_dir_path, job_name)
make_if_not_exist(snapshot_dir)
make_if_not_exist(log_path)
conf.model_path = snapshot_dir
conf.log_path = log_path
conf.job_name = job_name
return conf
def _load_model(self, model_path):
# define model
model_name = os.path.basename(model_path)
h_input, w_input, model_type, _ = 80, 80, None, 1
self.kernel_size = get_kernel(h_input, w_input,)
# self.model = MODEL_MAPPING["MiniFASNetV2"](conv6_kernel=self.kernel_size).to(self.device)
self.model = RepVggFTNet(32, [48, 64, 128], [4, 10, 8], num_out=2, num_classes=3).to(self.device)
# load model weight
state_dict = torch.load(model_path, map_location=self.device)
keys = iter(state_dict)
first_layer_name = keys.__next__()
if first_layer_name.find('module.') >= 0:
from collections import OrderedDict
new_state_dict = OrderedDict()
for key, value in state_dict.items():
# if str(key)[0:12]=="module.model":
name_key = key[7:]
new_state_dict[name_key] = value
self.model.load_state_dict(new_state_dict)
else:
self.model.load_state_dict(state_dict)
return None
def __init__(self):
super(AntiSpoofPredict, self).__init__()
self.num_class = 2
self.model_path = "/home/ubuntu/Silent-Face-Anti-Spoofing-master/saved_logs/snapshot/Anti_Spoofing_1_80-80/2021-01-11-13-19_Anti_Spoofing_1_80-80_model_iter-39500.pth"
model_name = os.path.basename(self.model_path)
h_input, w_input, model_type, _ = parse_model_name(model_name)
self.kernel_size = get_kernel(
h_input,
w_input,
)
self.model = MODEL_MAPPING["MiniFASNetV2SE"](
conv6_kernel=self.kernel_size, num_classes=self.num_class)
checkpoint = torch.load(self.model_path, map_location="cuda:3")
pretrain(self.model, checkpoint)
self.new_width = self.new_height = 224
self.transform = torchvision.transforms.Compose([
torchvision.transforms.Resize((self.new_width, self.new_height)),
torchvision.transforms.ToTensor(),
])
device = torch.device("cuda:3" if torch.cuda.is_available() else "cpu")
self.model.to(device)
self.model.eval()
def _load_model(self, model_path):
# define model
model_name = os.path.basename(model_path)
h_input, w_input, model_type, _ = parse_model_name(model_name)
self.kernel_size = get_kernel(
h_input,
w_input,
)
if model_type not in MODEL_MAPPING:
params = {
'embedding_size': 128,
'conv6_kernel': (5, 5),
'num_classes': 2,
'img_channel': 3,
'training': False
}
self.model = MultiFTNet(**params).to(self.device)
else:
self.model = MODEL_MAPPING[model_type](
conv6_kernel=self.kernel_size).to(self.device)
# load model weight
state_dict = torch.load(model_path, map_location=self.device)
keys = iter(state_dict)
first_layer_name = keys.__next__()
if first_layer_name.find('module.') >= 0:
from collections import OrderedDict
new_state_dict = OrderedDict()
for key, value in state_dict.items():
name_key = key[7:]
new_state_dict[name_key] = value
self.model.load_state_dict(new_state_dict)
else:
self.model.load_state_dict(state_dict)
return None
from lr_finder import LRFinder from src.model_lib.MultiFTNet import MultiFTNet from src.model_lib.MiniFASNet import MiniFASNetV1, MiniFASNetV2,MiniFASNetV1SE,MiniFASNetV2SE from src.utility import get_kernel from torch.nn import CrossEntropyLoss, MSELoss from torch import optim from src.data_io.dataset_loader import get_train_loader,get_eval_loader from src.default_config import get_default_config, update_config from train import parse_args import os os.environ["CUDA_VISIBLE_DEVICES"] = "3" kernel_size = get_kernel(80, 60) model = MultiFTNet(conv6_kernel = kernel_size) cls_criterion = CrossEntropyLoss() FT_criterion = MSELoss() from torch import optim # optimizer = optim.SGD(model.parameters(), # lr=0.1, # weight_decay=5e-4, # momentum=0.9) optimizer = optim.AdamW(model.parameters()) lr_finder = LRFinder(model, optimizer, cls_criterion,FT_criterion) conf = get_default_config() args = parse_args() conf = update_config(args, conf) trainloader = get_train_loader(conf) val_loader = get_eval_loader(conf) lr_finder.range_test(trainloader, end_lr=1, num_iter=100, step_mode="linear") lr_finder.plot(log_lr=False) lr_finder.reset()
