Method __init__

datasets/preprocessing/rio_preprocessing.py:19–54

(
        self,
        data_dir: str = "./data/raw/rio/rio",
        save_dir: str = "./data/processed/rio",
        modes: tuple = ("train", "validation", "test"),
        n_jobs: int = -1,
        git_repo: str = "./data/raw/rio/3RScan",
        label_db: str = "configs/scannet_preprocessing/label_database.yaml",
    )

class RioPreprocessing(BasePreprocessing):
    def __init__(
        self,
        data_dir: str = "./data/raw/rio/rio",
        save_dir: str = "./data/processed/rio",
        modes: tuple = ("train", "validation", "test"),
        n_jobs: int = -1,
        git_repo: str = "./data/raw/rio/3RScan",
        label_db: str = "configs/scannet_preprocessing/label_database.yaml",
    ):
        super().__init__(data_dir, save_dir, modes, n_jobs)

        git_repo = Path(git_repo)
        self.files = {}
        for mode in self.modes:
            mode = "val" if mode == "validation" else mode
            trainval_split_dir = git_repo / "splits"
            with open(Path(trainval_split_dir) / (mode + ".txt")) as f:
                # -1 because the last one is always empty
                split_file = f.read().split("\n")[:-1]

            filepaths = []
            for folder in split_file:
                filepaths.append(self.data_dir / folder / "mesh.refined.obj")
            mode = "validation" if mode == "val" else mode
            self.files[mode] = natsorted(filepaths)

        self.rio_to_scannet_label = {}
        with open(git_repo / "data" / "mapping.tsv") as f:
            reader = csv.reader(f, delimiter="\t")
            columns = next(reader)
            raw_category = columns.index("Label")
            nyu40class = columns.index("NYU40 Mapping")
            for row in reader:
                self.rio_to_scannet_label[row[raw_category]] = row[nyu40class]

        self.label_db = self._load_yaml(Path(label_db))

    def process_file(self, filepath, mode):
        """process_file.