A wrapper of repeated dataset. The length of repeated dataset will be `times` larger than the original dataset. This is useful when the data loading time is long but the dataset is small. Using RepeatDataset can reduce the data loading time between epochs. Args: dataset (
| 19 | |
| 20 | @DATASETS.register_module() |
| 21 | class RepeatDataset(object): |
| 22 | """A wrapper of repeated dataset. |
| 23 | The length of repeated dataset will be `times` larger than the original |
| 24 | dataset. This is useful when the data loading time is long but the dataset |
| 25 | is small. Using RepeatDataset can reduce the data loading time between |
| 26 | epochs. |
| 27 | Args: |
| 28 | dataset (:obj:`Dataset`): The dataset to be repeated. |
| 29 | times (int): Repeat times. |
| 30 | """ |
| 31 | |
| 32 | def __init__(self, dataset: Dataset, times: int): |
| 33 | self.dataset = dataset |
| 34 | self.times = times |
| 35 | |
| 36 | self._ori_len = len(self.dataset) |
| 37 | |
| 38 | def __getitem__(self, idx: int): |
| 39 | return self.dataset[idx % self._ori_len] |
| 40 | |
| 41 | def __len__(self): |
| 42 | return self.times * self._ori_len |