filename='detection-{epoch:02d}-{val_loss:.2f}',
save_top_k=10,
save_last=True,
verbose=True,
mode='min')
from dataset.dataset import get_train_transforms, get_valid_transforms, zaloDataset, collate_fn
train_dataset = zaloDataset(
root_path=os.path.join(
os.path.join(os.getcwd(),
'data/za_traffic_2020/traffic_train/images')),
file_name=os.path.join(
os.getcwd(),
'data/za_traffic_2020/traffic_train/train_traffic_sign_dataset.json'),
transforms=get_train_transforms())
valid_dataset = zaloDataset(
root_path=os.path.join(
os.path.join(os.getcwd(),
'data/za_traffic_2020/traffic_train/images')),
file_name=os.path.join(
os.getcwd(),
'data/za_traffic_2020/traffic_train/train_traffic_sign_dataset.json'),
transforms=get_valid_transforms())
train_loader = torch.utils.data.DataLoader(train_dataset,
batch_size=10,
shuffle=True,
num_workers=8,
collate_fn=collate_fn)
valid_loader = torch.utils.data.DataLoader(valid_dataset,
batch_size=10,
header=['epoch', 'loss', 'acc', 'lr'])
# criterion = nn.CrossEntropyLoss()
criterion = LabelSmoothing(smoothing=0.05).cuda(device_id)
# optimizer = optim.SGD(model.parameters(), lr=lr, momentum=0.9)
# optimizer = optim.Adam(model.parameters(), lr=lr)
optimizer = optim.AdamW(model.parameters(), lr=lr)
scheduler = optim.lr_scheduler.StepLR(optimizer, step_size=2, gamma=0.9)
is_train = False
if is_train:
xdl = DeeperForensicsDatasetNew(
real_npys=train_real_paths_npy,
fake_npys=train_fake_paths_npy,
is_one_hot=True,
transforms=get_train_transforms(size=300))
train_loader = DataLoader(xdl,
batch_size=batch_size,
shuffle=False,
num_workers=4,
sampler=BalanceClassSampler(
labels=xdl.get_labels(),
mode="downsampling"))
# train_loader = DataLoader(xdl, batch_size=batch_size, shuffle=True, num_workers=4)
train_dataset_len = len(xdl)
xdl_eval = DeeperForensicsDatasetNew(
real_npys=val_real_paths_npy,
fake_npys=val_fake_paths_npy,
is_one_hot=False,
transforms=get_valid_transforms(size=300))