def get_model(config):
if config.MODEL.NAME == 'hrnetv2':
model = get_hrnetv2()
print('model: hrnetv2')
elif config.MODEL.NAME == 'resnet50_upernet':
model = get_resnet50_upernet()
print('model: resnet50_upernet')
elif config.MODEL.NAME == 'resnet101_upernet':
model = get_resnet101_upernet()
print('model: resnet101_upernet')
elif config.MODEL.NAME == 'acnet':
model = ACNet(num_class=4, pretrained=True)
print('model: acnet')
elif config.MODEL.NAME == 'deeplabv3':
model = get_deeplabv3()
print('model: deeplabv3')
elif config.MODEL.NAME == 'deeplab_xception':
model = DeepLab(backbone='xception',
output_stride=16,
num_classes=4,
sync_bn=False,
freeze_bn=False)
else:
model_architecture = config.MODEL.ARCHITECTURE
model_encoder = config.MODEL.ENCODER
model_pretrained = config.MODEL.PRETRAINED
if model_architecture == 'Unet':
model = Unet(model_encoder,
encoder_weights=model_pretrained,
classes=4,
attention_type='scse')
elif model_architecture == 'Linknet':
model = Linknet(model_encoder,
encoder_weights=model_pretrained,
classes=4)
elif model_architecture == 'FPN' or model_architecture == 'PSPNet':
model = FPN(model_encoder,
encoder_weights=model_pretrained,
classes=4)
print('architecture:', model_architecture, 'encoder:', model_encoder,
'pretrained on:', model_pretrained)
if config.PARALLEL:
model = nn.DataParallel(model)
print('[*] num parameters:', count_parameters(model))
return model
def get_model(config):
model_architecture = config.ARCHITECTURE
model_encoder = config.ENCODER
# activation은 eval 모드일 때 적용해 주는 거라 train 때에는 직접 sigmoid 쳐야한다.
if model_architecture == 'Unet':
model = Unet(model_encoder, encoder_weights='imagenet', classes=4, attention_type='scse')
elif model_architecture == 'FPN':
model = FPN(model_encoder, encoder_weights='imagenet', classes=4)
print('architecture:', model_architecture, 'encoder:', model_encoder)
return model
def fpn(backbone, pretrained_weights=None, classes=1, activation='sigmoid'):
device = torch.device("cuda")
model = FPN(encoder_name=backbone,
encoder_weights=pretrained_weights,
classes=classes,
activation=activation)
model.to(device)
model.eval() # 위치 확인해볼것
return model
limit_val_num_samples=100 if debug else None)
accumulation_steps = 2
prepare_batch = prepare_batch_fp32
# Image denormalization function to plot predictions with images
def img_denormalize(nimg):
img = denormalize(nimg, mean=mean, std=std)
return img[(0, 1, 2), :, :]
#################### Model ####################
model = FPN(in_channels=5, encoder_name='se_resnext50_32x4d', classes=2)
#################### Solver ####################
num_epochs = 75
criterion = nn.CrossEntropyLoss(weight=torch.tensor([0.62, 1.45]))
lr = 0.001
weight_decay = 1e-4
optimizer = optim.Adam(model.parameters(), lr=1.0, weight_decay=weight_decay)
le = len(train_loader)
def lambda_lr_scheduler(iteration, lr0, n, a):
[A.Normalize(mean=mean, std=std, max_pixel_value=max_value),
ToTensorV2()])
_, data_loader, _ = get_train_val_loaders(
train_ds,
val_ds,
train_transforms=transforms,
val_transforms=transforms,
batch_size=batch_size,
num_workers=num_workers,
val_batch_size=batch_size,
pin_memory=True,
)
prepare_batch = inference_prepare_batch_f32
# Image denormalization function to plot predictions with images
img_denormalize = partial(denormalize, mean=mean, std=std)
#################### Model ####################
model = FPN(encoder_name='se_resnext50_32x4d', classes=2, encoder_weights=None)
run_uuid = "5230c20f609646cb9870a211036ea5cb"
weights_filename = "best_model_67_val_miou_bg=0.7574240313552584.pth"
has_targets = True
tta_transforms = tta.Compose([
tta.Rotate90(angles=[90, -90, 180]),
])
val_batch_size=val_batch_size,
pin_memory=True,
train_sampler=train_sampler,
limit_train_num_samples=100 if debug else None,
limit_val_num_samples=100 if debug else None)
accumulation_steps = 3
prepare_batch = prepare_batch_fp32
# Image denormalization function to plot predictions with images
img_denormalize = partial(denormalize, mean=mean, std=std)
#################### Model ####################
model = FPN(encoder_name='efficientnet-b4', classes=num_classes)
#################### Solver ####################
num_epochs = 75
criterion = nn.CrossEntropyLoss(weight=torch.tensor([0.75, 1.5]))
lr = 0.001
weight_decay = 1e-4
optimizer = optim.Adam(model.parameters(), lr=1.0, weight_decay=weight_decay)
le = len(train_loader)
def lambda_lr_scheduler(iteration, lr0, n, a):
val_batch_size=val_batch_size,
pin_memory=True,
train_sampler=train_sampler,
limit_train_num_samples=100 if debug else None,
limit_val_num_samples=100 if debug else None)
accumulation_steps = 2
prepare_batch = prepare_batch_fp32
# Image denormalization function to plot predictions with images
img_denormalize = partial(denormalize, mean=mean, std=std)
#################### Model ####################
model = FPN(encoder_name='se_resnext50_32x4d', classes=num_classes)
#################### Solver ####################
num_epochs = 75
criterion = nn.CrossEntropyLoss(weight=torch.tensor([0.75, 1.5]))
lr = 0.001
weight_decay = 1e-4
optimizer = optim.Adam(model.parameters(), lr=1.0, weight_decay=weight_decay)
le = len(train_loader)
def lambda_lr_scheduler(iteration, lr0, n, a):
def load_model_fpn(_model_weights, is_inference=False):
print("Using weights {}".format(_model_weights))
if _model_weights == "imagenet":
model = FPN(unet_encoder,
encoder_weights="imagenet",
classes=4,
activation=None)
if is_inference:
model.eval()
return model
else:
model = FPN(unet_encoder,
encoder_weights=None,
classes=4,
activation=None)
if is_inference:
model.eval()
if _model_weights is not None:
device = torch.device("cuda")
model.to(device)
state = torch.load(
_model_weights) # , map_location=lambda storage, loc: storage)
model.load_state_dict(state["state_dict"])
# new_state_dict = OrderedDict()
#
# for k, v in state['state_dict'].items():
# if k in model.state_dict():
# new_state_dict[k] = v
# model = model.load_state_dict(new_state_dict)
return model
def __init__(self):
super(HuBMAPModel, self).__init__()
self.model = FPN(encoder_name = ENCODER_NAME,
encoder_weights = 'imagenet',
classes = 1,
activation = None)
unet_resnet34 = unet_resnet34.eval()
device = torch.device("cuda")
model_senet = Unet('se_resnext50_32x4d',
encoder_weights=None,
classes=4,
activation=None)
model_senet.to(device)
model_senet.eval()
state = torch.load(
'../input/senetmodels/senext50_30_epochs_high_threshold.pth',
map_location=lambda storage, loc: storage)
model_senet.load_state_dict(state["state_dict"])
model_fpn91lb = FPN(encoder_name="se_resnext50_32x4d",
classes=4,
activation=None,
encoder_weights=None)
model_fpn91lb.to(device)
model_fpn91lb.eval()
#state = torch.load('../input/fpnseresnext/model_se_resnext50_32x4d_fold_0_epoch_7_dice_0.935771107673645.pth', map_location=lambda storage, loc: storage)
state = torch.load(
'../input/fpnse50dice944/model_se_resnext50_32x4d_fold_0_epoch_26_dice_0.94392.pth',
map_location=lambda storage, loc: storage)
model_fpn91lb.load_state_dict(state["state_dict"])
model_fpn91lb_pseudo = FPN(encoder_name="se_resnext50_32x4d",
classes=4,
activation=None,
encoder_weights=None)
model_fpn91lb_pseudo.to(device)
model_fpn91lb_pseudo.eval()
def predict_valid():
inputdir = "/data/Thoracic_OAR/"
transform = valid_aug(image_size=512)
# nii_files = glob.glob(inputdir + "/*/data.nii.gz")
folds = [0, 1, 2, 3, 4]
for fold in folds:
print(fold)
outdir = f"/data/Thoracic_OAR_predict/FPN-seresnext50/"
log_dir = f"/logs/ss_miccai/FPN-se_resnext50_32x4d-fold-{fold}"
# model = VNet(
# encoder_name='se_resnext50_32x4d',
# encoder_weights=None,
# classes=7,
# # activation='sigmoid',
# group_norm=False,
# center='none',
# attention_type='scse',
# reslink=True,
# multi_task=False
# )
model = FPN(encoder_name='se_resnext50_32x4d',
encoder_weights=None,
classes=7)
ckp = os.path.join(log_dir, "checkpoints/best.pth")
checkpoint = torch.load(ckp)
model.load_state_dict(checkpoint['model_state_dict'])
model = nn.DataParallel(model)
model = model.to(device)
df = pd.read_csv(f'./csv/5folds/valid_{fold}.csv')
patient_ids = df.patient_id.unique()
for patient_id in patient_ids:
print(patient_id)
nii_file = f"{inputdir}/{patient_id}/data.nii.gz"
image_slices, ct_image = extract_slice(nii_file)
dataset = TestDataset(image_slices, transform)
dataloader = DataLoader(dataset=dataset,
num_workers=4,
batch_size=8,
drop_last=False)
pred_mask, pred_logits = predict(model, dataloader)
# import pdb
# pdb.set_trace()
pred_mask = np.argmax(pred_mask, axis=1).astype(np.uint8)
pred_mask = SimpleITK.GetImageFromArray(pred_mask)
pred_mask.SetDirection(ct_image.GetDirection())
pred_mask.SetOrigin(ct_image.GetOrigin())
pred_mask.SetSpacing(ct_image.GetSpacing())
# patient_id = nii_file.split("/")[-2]
patient_dir = f"{outdir}/{patient_id}"
os.makedirs(patient_dir, exist_ok=True)
patient_pred = f"{patient_dir}/predict.nii.gz"
SimpleITK.WriteImage(pred_mask, patient_pred)
val_batch_size=val_batch_size,
pin_memory=True,
train_sampler=train_sampler,
limit_train_num_samples=100 if debug else None,
limit_val_num_samples=100 if debug else None)
accumulation_steps = 2
prepare_batch = prepare_batch_fp32
# Image denormalization function to plot predictions with images
img_denormalize = partial(denormalize, mean=mean, std=std)
#################### Model ####################
model = FPN(encoder_name='dpn92', classes=num_classes)
#################### Solver ####################
num_epochs = 75
criterion = nn.CrossEntropyLoss(weight=torch.tensor([0.75, 1.5]))
lr = 0.001
weight_decay = 1e-4
optimizer = optim.Adam(model.parameters(), lr=1.0, weight_decay=weight_decay)
le = len(train_loader)
def lambda_lr_scheduler(iteration, lr0, n, a):