print(largest_object)
target_tensor = data.target_to_tensor(largest_object)
# Expected : {'bboxes': torch.tensor([0.5240, 0.5735, 0.8360, 0.7534]), 'labels': torch.tensor([5])})
print(target_tensor)
# The datasets is already downloaded on the cluster
dataset_dir = "/opt/Datasets/Pascal-VOC2012/"
download = False
# How do we preprocessing the image (e.g. none, crop, shrink)
image_transform_params = {'image_mode': 'none'}
# How do we preprocess the targets
target_transform_params = {
'target_mode': 'largest_bbox',
'image_transform_params': image_transform_params
}
# The post-processing of the image
image_transform = transforms.ToTensor()
train_dataset, valid_dataset = data.make_trainval_dataset(
dataset_dir=dataset_dir,
image_transform_params=image_transform_params,
transform=image_transform,
target_transform_params=target_transform_params,
download=download)
print(train_dataset[203])
}
}
target_transform_params = {
'target_mode': 'only_cls',
'image_transform_params': image_transform_params
}
imagenet_preprocessing = transforms.Normalize(mean=[0.485, 0.456, 0.406],
std=[0.229, 0.224, 0.225])
image_transform = transforms.Compose(
[transforms.ToTensor(), imagenet_preprocessing])
train_dataset, val_dataset = data.make_trainval_dataset(
image_transform_params=image_transform_params,
transform=image_transform,
target_transform_params=target_transform_params,
download=True)
print(train_dataset[0])
train_loader = torch.utils.data.DataLoader(dataset=train_dataset,
batch_size=32,
shuffle=True,
**kwargs)
valid_loader = torch.utils.data.DataLoader(dataset=val_dataset,
batch_size=32,
shuffle=False,
**kwargs)
############################################ Model
model = extract.FeatureExtractor()