datamodules.RotNet.datasets package
Submodules
datamodules.RotNet.datasets.cropped_dataset module
Load a dataset of historic documents by specifying the folder where its located.
- class CroppedRotNet(path: Path, data_folder_name: str, gt_folder_name: Optional[str] = None, selection: Optional[Union[int, List[str]]] = None, is_test: bool = False, image_transform: Optional[callable] = None)[source]
Bases:
CroppedHisDBDatasetDataset implementation of the RotNet paper of Gidaris et al.. This dataset is used for the DivaHisDB dataset in a cropped setup.
The structure of the folder should be as follows:
data_dir ├── train_folder_name │ ├── data_folder_name │ │ ├── original_image_name_1 │ │ │ ├── image_crop_1.png │ │ │ ├── ... │ │ │ └── image_crop_N.png │ │ └──original_image_name_N │ │ ├── image_crop_1.png │ │ ├── ... │ │ └── image_crop_N.png │ └── gt_folder_name │ ├── original_image_name_1 │ │ ├── image_crop_1.png │ │ ├── ... │ │ └── image_crop_N.png │ └──original_image_name_N │ ├── image_crop_1.png │ ├── ... │ └── image_crop_N.png ├── validation_folder_name │ ├── data_folder_name │ │ ├── original_image_name_1 │ │ │ ├── image_crop_1.png │ │ │ ├── ... │ │ │ └── image_crop_N.png │ │ └──original_image_name_N │ │ ├── image_crop_1.png │ │ ├── ... │ │ └── image_crop_N.png │ └── gt_folder_name │ ├── original_image_name_1 │ │ ├── image_crop_1.png │ │ ├── ... │ │ └── image_crop_N.png │ └──original_image_name_N │ ├── image_crop_1.png │ ├── ... │ └── image_crop_N.png └── test_folder_name ├── data_folder_name │ ├── original_image_name_1 │ │ ├── image_crop_1.png │ │ ├── ... │ │ └── image_crop_N.png │ └──original_image_name_N │ ├── image_crop_1.png │ ├── ... │ └── image_crop_N.png └── gt_folder_name ├── original_image_name_1 │ ├── image_crop_1.png │ ├── ... │ └── image_crop_N.png └──original_image_name_N ├── image_crop_1.png ├── ... └── image_crop_N.png- Parameters:
path (Path) – Path to root dir of the dataset (folder containing the train/val/test folder)
data_folder_name (str) – Name of the folder containing the train/val/test folder
gt_folder_name (str) – Name of the folder containing the train/val/test folder
selection (Union[int, List[str]]) – If you only want to use a subset of the dataset, you can specify the name of the files (without the file extension) in a list. If you want to use all files, set this parameter to None.
is_test (bool) – If True, the it returns additional information that are important for the test set.
image_transform –
- static get_gt_data_paths(directory: Path, data_folder_name: str, gt_folder_name: Optional[str] = None, selection: Optional[Union[int, List[str]]] = None) List[Path][source]
Creates the list of paths to the original images.
Structure of the folder:
dictionary ├── data_folder_name │ ├── original_image_name_1 │ │ ├── image_crop_1.png │ │ ├── ... │ │ └── image_crop_N.png │ └──original_image_name_N │ ├── image_crop_1.png │ ├── ... │ └── image_crop_N.png └── gt_folder_name ├── original_image_name_1 │ ├── image_crop_1.png │ ├── ... │ └── image_crop_N.png └──original_image_name_N ├── image_crop_1.png ├── ... └── image_crop_N.png- Parameters:
directory (Path) – Path to root dir of split
data_folder_name (str) – Name of the folder containing the data
gt_folder_name (str) – Name of the folder containing the ground truth
selection (Union[int, List[str]]) – If you only want to use a subset of the dataset, you can specify the name of the files (without the file extension) in a list. If you want to use all files, set this parameter to None.
- Returns:
List of paths to the original images
- Return type:
List[Path]