batch_size = 4
num_workers = 12
mean = [-17.398721187929123, -10.020421713800838, -12.10841437771272]
std = [6.290316422115964, 5.776936185931195, 5.795418280085563]
max_value = 1.0
transforms = A.Compose(
[A.Normalize(mean=mean, std=std, max_pixel_value=max_value),
ToTensorV2()])
data_loader = get_inference_dataloader(
test_dataset,
transforms=transforms,
batch_size=batch_size,
num_workers=num_workers,
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 = "30187583292246f6999d499642372da9"
weights_filename = "best_model_43_val_miou_bg=0.7530081542328186.pth"
train_ds,
val_ds,
train_transforms=train_transforms,
val_transforms=val_transforms,
batch_size=batch_size // 2,
num_workers=num_workers // 2,
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)
unsup_train_loader = get_inference_dataloader(
test_dataset,
transforms=train_transforms,
batch_size=batch_size // 2,
num_workers=num_workers // 2,
pin_memory=True,
)
# 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 = LWRefineNet(num_channels=3, num_classes=num_classes)