def compute_time():
"""
计算１00次，取平均值
:return:
"""
model_googlenet_bn = googlenet_bn.GoogLeNet_BN(num_classes=1000)
model_googlenet = googlenet.GoogLeNet(num_classes=1000)
model_googlenet_bn.eval()
model_googlenet.eval()
total_time_googlenet_bn = 0.0
total_time_googlenet = 0.0
epoch = 100
for i in range(epoch):
data = torch.randn((1, 3, 224, 224))
start = time.time()
outputs = model_googlenet_bn.forward(data)
end = time.time()
total_time_googlenet_bn += (end - start)
start = time.time()
outputs = model_googlenet.forward(data)
end = time.time()
total_time_googlenet += (end - start)
print('[googlenet_bn] time: {:.4f}'.format(total_time_googlenet_bn /
epoch))
print('[googlenet] time: {:.4f}'.format(total_time_googlenet / epoch))
print('time_googlenet / time_googlenet_bn: {:.3f}'.format(
total_time_googlenet / total_time_googlenet_bn))
def compute_param():
model_googlenet_bn = googlenet_bn.GoogLeNet_BN(num_classes=1000)
model_googlenet = googlenet.GoogLeNet(num_classes=1000)
model_googlenet_bn.eval()
model_googlenet.eval()
num_googlenet_bn = util.num_model(model_googlenet_bn)
num_googlenet = util.num_model(model_googlenet)
print('[googlenet_bn] param num: {}'.format(num_googlenet_bn))
print('[googlenet] param num: {}'.format(num_googlenet))
print('num_googlenet_bn / num_googlenet: {:.2f}'.format(num_googlenet_bn /
num_googlenet))
def compute_param():
model_alexnet = AlexNet(num_classes=1000)
model_googlenet = googlenet.GoogLeNet(num_classes=1000)
model_alexnet.eval()
model_googlenet.eval()
num_alexnet = util.num_model(model_alexnet)
num_googlenet = util.num_model(model_googlenet)
print('[alexnet] param num: {}'.format(num_alexnet))
print('[googlenet] param num: {}'.format(num_googlenet))
print('num_alexnet / num_googlenet: {:.2f}'.format(num_alexnet /
num_googlenet))
############ choose loss and backbone ######################
if opt.loss == 'focal_loss':
criterion = focal_loss.FocalLoss(gamma=2)
else:
criterion = torch.nn.CrossEntropyLoss()
if opt.backbone == 'resnet18':
model = resnet.resnet_face18(use_se=opt.use_se)
elif opt.backbone == 'resnet50':
model = resnet.resnet50()
elif opt.backbone == 'resnet101':
model = resnet.resnet101()
elif opt.backbone == 'resnet152':
model = resnet.resnet152()
elif opt.backbone == 'googlenet':
model = googlenet.GoogLeNet()
if opt.metric == 'add_margin':
metric_fc = metrics.AddMarginProduct(1024,
opt.num_classes,
s=30,
m=0.35)
elif opt.metric == 'arc_margin':
metric_fc = metrics.ArcMarginProduct(512,
opt.num_classes,
s=30,
m=0.5,
easy_margin=opt.easy_margin)
elif opt.metric == 'sphere':
metric_fc = metrics.SphereProduct(512, opt.num_classes, m=4)
else:
if __name__ == '__main__':
device = util.get_device()
# device = 'cpu'
data_loaders, data_sizes = load_data('./data/pascal-voc')
print(data_loaders)
print(data_sizes)
res_loss = dict()
res_acc = dict()
for name in ['googlenet_bn', 'googlenet']:
if name == 'googlenet_bn':
model = googlenet_bn.GoogLeNet_BN(num_classes=20)
else:
model = googlenet.GoogLeNet(num_classes=20)
model.eval()
# print(model)
model = model.to(device)
criterion = nn.CrossEntropyLoss()
optimizer = optim.Adam(model.parameters(), lr=1e-3)
lr_schduler = optim.lr_scheduler.StepLR(optimizer,
step_size=8,
gamma=0.96)
util.check_dir('./data/models/')
best_model, loss_dict, acc_dict = train_model(data_loaders,
data_sizes,
name,
model,
net = efficientnet.CEfficientNet(num_classes=num_classes,
pretrained=args.pretrain,
model_name=args.model)
elif args.model == "efficientnet-b4":
net = efficientnet.CEfficientNet(num_classes=num_classes,
pretrained=args.pretrain,
model_name=args.model)
elif args.model == "efficientnet-b5":
net = efficientnet.CEfficientNet(num_classes=num_classes,
pretrained=args.pretrain,
model_name=args.model)
elif args.model == "resnext50":
net = resnext.resnext50(num_classes=num_classes)
elif args.model == "googlenet":
net = googlenet.GoogLeNet(num_classes=num_classes)
elif args.model == "densenet":
net = densenet.DenseNet_CIFAR(num_classes=num_classes)
else:
raise "please check model"
# freeze
# count = 0
# for param in net.parameters():
# count += 1
# for i, param in enumerate(net.parameters()):
# if i <= count-1 - 10:
# param.requires_grad = False
str_pretrain = ""
if args.pretrain:
def main():
parser = argparse.ArgumentParser(description='Speech Emotion Recognition')
parser.add_argument('--hidden_size',
type=int,
default=512,
help='hidden size of model (default: 256)')
parser.add_argument('--layer_size',
type=int,
default=3,
help='number of layers of model (default: 3)')
parser.add_argument('--n_class',
type=int,
default=7,
help='number of classes of data (default: 7)')
parser.add_argument('--dropout',
type=float,
default=0.2,
help='dropout rate in training (default: 0.2')
parser.add_argument('--bidirectional',
default=True,
action='store_true',
help='use bidirectional RNN (default: False')
parser.add_argument('--batch_size',
type=int,
default=4,
help='batch size in training (default: 32')
parser.add_argument(
'--workers',
type=int,
default=4,
help='number of workers in dataset loader (default: 4)')
parser.add_argument('--max_epochs',
type=int,
default=30,
help='number of max epochs in training (default: 10)')
parser.add_argument('--lr',
type=float,
default=1e-04,
help='learning rate (default: 0.0001)')
parser.add_argument('--no_cuda',
action='store_true',
default=False,
help='disables CUDA training')
parser.add_argument('--seed',
type=int,
default=1,
help='random seed (default: 1)')
parser.add_argument('--save_name',
type=str,
default='model',
help='the name of model')
parser.add_argument('--mode', type=str, default='train')
args = parser.parse_args()
random.seed(args.seed)
torch.manual_seed(args.seed)
torch.cuda.manual_seed_all(args.seed)
args.cuda = not args.no_cuda and torch.cuda.is_available()
device = torch.device('cuda' if args.cuda else 'cpu')
# N_FFT: defined in loader.py
feature_size = N_FFT / 2 + 1
cnn = googlenet.GoogLeNet(feature_size=feature_size)
rnn = RNN.RNN(cnn.feature_size,
args.hidden_size,
args.n_class,
input_dropout_p=args.dropout,
dropout_p=args.dropout,
n_layers=args.layer_size,
bidirectional=args.bidirectional,
rnn_cell='gru',
variable_lengths=False)
model = CRNN.CRNN(cnn, rnn)
model.flatten_parameters()
model = nn.DataParallel(model).to(device)
optimizer = optim.Adam(model.module.parameters(), lr=args.lr)
criterion = nn.CrossEntropyLoss(reduction='sum').to(device)
if args.mode != 'train':
return
data_download()
wav_paths = [
os.path.join('./dataset/wav', fname)
for fname in os.listdir('./dataset/wav')
]
best_acc = 0
begin_epoch = 0
loss_acc = [[], [], [], []]
train_batch_num, train_dataset_list, valid_dataset, test_dataset = split_dataset(
args, wav_paths, dataset_ratio=[0.7, 0.1, 0.2])
logger.info('start')
train_begin = time.time()
for epoch in range(begin_epoch, args.max_epochs):
train_queue = queue.Queue(args.workers * 2)
train_loader = MultiLoader(train_dataset_list, train_queue,
args.batch_size, args.workers)
train_loader.start()
train_loss, train_acc = train(model, train_batch_num, train_queue,
criterion, optimizer, device,
train_begin, args.workers, 10)
logger.info('Epoch %d (Training) Loss %0.4f Acc %0.4f' %
(epoch, train_loss, train_acc))
train_loader.join()
loss_acc[0].append(train_loss)
loss_acc[1].append(train_acc)
valid_queue = queue.Queue(args.workers * 2)
valid_loader = BaseDataLoader(valid_dataset, valid_queue,
args.batch_size, 0)
valid_loader.start()
eval_loss, eval_acc = evaluate(model, valid_loader, valid_queue,
criterion, device)
logger.info('Epoch %d (Evaluate) Loss %0.4f Acc %0.4f' %
(epoch, eval_loss, eval_acc))
valid_loader.join()
loss_acc[2].append(eval_loss)
loss_acc[3].append(eval_acc)
best_model = (eval_acc > best_acc)
if best_model:
best_acc = eval_acc
torch.save(model.state_dict(), './save_model/best_model.pt')
save_epoch = epoch
model.load_state_dict(torch.load('./save_model/best_model.pt'))
test_queue = queue.Queue(args.workers * 2)
test_loader = BaseDataLoader(test_dataset, test_queue, args.batch_size, 0)
test_loader.start()
test_loss, test_acc = evaluate(model, test_loader, test_queue, criterion,
device)
logger.info('Epoch %d (Test) Loss %0.4f Acc %0.4f' %
(save_epoch, test_loss, test_acc))
test_loader.join()
save_data(loss_acc, test_loss, test_acc)
plot_data(loss_acc, test_loss, test_acc)
return 0
def main(args):
BATCH_SIZE = args.batch_size
LR = args.learning_rate
EPOCH = args.epoch
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
use_gpu = torch.cuda.is_available()
data_transforms = {
transforms.Compose([
transforms.Resize(320),
transforms.CenterCrop(299),
transforms.ToTensor(),
transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5))
])
}
transform = transforms.Compose([
transforms.Resize(size=(227, 227)),
transforms.RandomRotation(20),
transforms.RandomHorizontalFlip(),
transforms.ToTensor(), # 将图片转换为Tensor,归一化至[0,1]
transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5))
])
train_dataset = torchvision.datasets.ImageFolder(root=args.train_images,
transform=transform)
train_loader = torch.utils.data.DataLoader(train_dataset,
batch_size=BATCH_SIZE,
shuffle=True)
# 从文件夹中读取validation数据
validation_dataset = torchvision.datasets.ImageFolder(
root=args.test_images, transform=transform)
print(validation_dataset.class_to_idx)
test_loader = torch.utils.data.DataLoader(validation_dataset,
batch_size=BATCH_SIZE,
shuffle=True)
if args.model_name == "densenet":
Net = densenet.DenseNet().to(device)
if args.model_name == "alexnet":
Net = alexnet.AlexNet().to(device)
if args.model_name == "googlenet":
Net = googlenet.GoogLeNet().to(device)
if args.model_name == "mobilenet":
Net = mobilenet.MobileNetV2().to(device)
if args.model_name == "mnasnet":
Net = mnasnet.mnasnet1_0().to(device)
if args.model_name == "squeezenet":
Net = squeezenet.SqueezeNet().to(device)
if args.model_name == "resnet":
Net = resnet.resnet50().to(device)
if args.model_name == "vgg":
Net = vgg.vgg19().to(device)
if args.model_name == "shufflenetv2":
Net = shufflenetv2.shufflenet_v2_x1_0().to(device)
criterion = nn.CrossEntropyLoss()
opti = torch.optim.Adam(Net.parameters(), lr=LR)
if __name__ == '__main__':
Accuracy_list = []
Loss_list = []
for epoch in range(EPOCH):
sum_loss = 0.0
correct1 = 0
total1 = 0
for i, (images, labels) in enumerate(train_loader):
num_images = images.size(0)
images = Variable(images.to(device))
labels = Variable(labels.to(device))
if args.model_name == 'googlenet':
out = Net(images)
out = out[0]
else:
out = Net(images)
_, predicted = torch.max(out.data, 1)
total1 += labels.size(0)
correct1 += (predicted == labels).sum().item()
loss = criterion(out, labels)
print(loss)
opti.zero_grad()
loss.backward()
opti.step()
# 每训练100个batch打印一次平均loss
sum_loss += loss.item()
if i % 10 == 9:
print('train loss [%d, %d] loss: %.03f' %
(epoch + 1, i + 1, sum_loss / 2000))
print("train acc %.03f" % (100.0 * correct1 / total1))
sum_loss = 0.0
Accuracy_list.append(100.0 * correct1 / total1)
print('accurary={}'.format(100.0 * correct1 / total1))
Loss_list.append(loss.item())
x1 = range(0, EPOCH)
x2 = range(0, EPOCH)
y1 = Accuracy_list
y2 = Loss_list
total_test = 0
correct_test = 0
for i, (images, labels) in enumerate(test_loader):
start_time = time.time()
print('time_start', start_time)
num_images = images.size(0)
print('num_images', num_images)
images = Variable(images.to(device))
labels = Variable(labels.to(device))
print("GroundTruth", labels)
if args.model_name == 'googlenet':
out = Net(images)[0]
out = out[0]
else:
out = Net(images)
_, predicted = torch.max(out.data, 1)
print("predicted", predicted)
correct_test += (predicted == labels).sum().item()
total_test += labels.size(0)
print('time_usage', (time.time() - start_time) / args.batch_size)
print('total_test', total_test)
print('correct_test', correct_test)
print('accurary={}'.format(100.0 * correct_test / total_test))
plt.subplot(2, 1, 1)
plt.plot(x1, y1, 'o-')
plt.title('Train accuracy vs. epoches')
plt.ylabel('Train accuracy')
plt.subplot(2, 1, 2)
plt.plot(x2, y2, '.-')
plt.xlabel('Train loss vs. epoches')
plt.ylabel('Train loss')
# plt.savefig("accuracy_epoch" + str(EPOCH) + ".png")
plt.savefig(args.output_dir + '/' + 'accuracy_epoch' + str(EPOCH) +
'.png')
plt.show()
torch.save(args.output_dir, args.model_name + '.pth')