diff options
| -rw-r--r-- | README.md | 3 | ||||
| -rw-r--r-- | text_recognizer/data/iam_extended_paragraphs.py | 6 | ||||
| -rw-r--r-- | text_recognizer/data/iam_paragraphs.py | 9 | ||||
| -rw-r--r-- | text_recognizer/models/base.py | 3 | ||||
| -rw-r--r-- | text_recognizer/models/metrics.py | 4 | ||||
| -rw-r--r-- | text_recognizer/models/transformer.py | 1 |
6 files changed, 12 insertions, 14 deletions
@@ -26,7 +26,7 @@ python build-transitions --tokens iamdb_1kwp_tokens_1000.txt --lexicon iamdb_1kw (TODO: Not working atm, needed for GTN loss function) ## Todo -- [ ] Efficient-net b0 + transformer decoder +- [x] Efficient-net b0 + transformer decoder - [ ] Load everything with hydra, get it to work - [ ] Tests - [ ] Evaluation @@ -34,6 +34,7 @@ python build-transitions --tokens iamdb_1kwp_tokens_1000.txt --lexicon iamdb_1kw - [ ] attr refactor - [ ] Refactor once more - [ ] fix linting +- [ ] fix loading of transform iam paragraph ## Run Sweeps diff --git a/text_recognizer/data/iam_extended_paragraphs.py b/text_recognizer/data/iam_extended_paragraphs.py index 23e424d..0e97801 100644 --- a/text_recognizer/data/iam_extended_paragraphs.py +++ b/text_recognizer/data/iam_extended_paragraphs.py @@ -16,6 +16,7 @@ class IAMExtendedParagraphs(BaseDataModule): augment: bool = attr.ib(default=True) train_fraction: float = attr.ib(default=0.8) word_pieces: bool = attr.ib(default=False) + num_classes: int = attr.ib(init=False) def __attrs_post_init__(self) -> None: self.iam_paragraphs = IAMParagraphs( @@ -35,8 +36,7 @@ class IAMExtendedParagraphs(BaseDataModule): self.dims = self.iam_paragraphs.dims self.output_dims = self.iam_paragraphs.output_dims - self.mapping = self.iam_paragraphs.mapping - self.inverse_mapping = self.iam_paragraphs.inverse_mapping + self.num_classes = self.iam_paragraphs.num_classes def prepare_data(self) -> None: """Prepares the paragraphs data.""" @@ -58,7 +58,7 @@ class IAMExtendedParagraphs(BaseDataModule): """Returns info about the dataset.""" basic = ( "IAM Original and Synthetic Paragraphs Dataset\n" # pylint: disable=no-member - f"Num classes: {len(self.mapping)}\n" + f"Num classes: {len(self.num_classes)}\n" f"Dims: {self.dims}\n" f"Output dims: {self.output_dims}\n" ) diff --git a/text_recognizer/data/iam_paragraphs.py b/text_recognizer/data/iam_paragraphs.py index 82058e0..7ba1077 100644 --- a/text_recognizer/data/iam_paragraphs.py +++ b/text_recognizer/data/iam_paragraphs.py @@ -38,20 +38,17 @@ MAX_LABEL_LENGTH = 682 class IAMParagraphs(BaseDataModule): """IAM handwriting database paragraphs.""" + num_classes: int = attr.ib() augment: bool = attr.ib(default=True) train_fraction: float = attr.ib(default=0.8) - word_pieces: bool = attr.ib(default=False) dims: Tuple[int, int, int] = attr.ib( init=False, default=(1, IMAGE_HEIGHT, IMAGE_WIDTH) ) output_dims: Tuple[int, int] = attr.ib(init=False, default=(MAX_LABEL_LENGTH, 1)) + inverse_mapping: Dict[str, int] = attr.ib(init=False) def __attrs_post_init__(self) -> None: - self.mapping, self.inverse_mapping, _ = emnist_mapping( - extra_symbols=[NEW_LINE_TOKEN] - ) - if self.word_pieces: - self.mapping = WordPieceMapping() + _, self.inverse_mapping, _ = emnist_mapping(extra_symbols=[NEW_LINE_TOKEN]) def prepare_data(self) -> None: """Create data for training/testing.""" diff --git a/text_recognizer/models/base.py b/text_recognizer/models/base.py index dfb4ca4..caf63c1 100644 --- a/text_recognizer/models/base.py +++ b/text_recognizer/models/base.py @@ -23,12 +23,9 @@ class BaseLitModel(LightningModule): criterion_config: DictConfig = attr.ib(converter=DictConfig) optimizer_config: DictConfig = attr.ib(converter=DictConfig) lr_scheduler_config: DictConfig = attr.ib(converter=DictConfig) - interval: str = attr.ib() monitor: str = attr.ib(default="val/loss") - loss_fn: Type[nn.Module] = attr.ib(init=False) - train_acc: torchmetrics.Accuracy = attr.ib( init=False, default=torchmetrics.Accuracy() ) diff --git a/text_recognizer/models/metrics.py b/text_recognizer/models/metrics.py index 9793157..0eb42dc 100644 --- a/text_recognizer/models/metrics.py +++ b/text_recognizer/models/metrics.py @@ -26,7 +26,9 @@ class CharacterErrorRate(Metric): bsz = preds.shape[0] for index in range(bsz): pred = [p for p in preds[index].tolist() if p not in self.ignore_indices] - target = [t for t in targets[index].tolist() if t not in self.ignore_indices] + target = [ + t for t in targets[index].tolist() if t not in self.ignore_indices + ] distance = editdistance.distance(pred, target) error = distance / max(len(pred), len(target)) self.error += error diff --git a/text_recognizer/models/transformer.py b/text_recognizer/models/transformer.py index 7a9d566..0e01bb5 100644 --- a/text_recognizer/models/transformer.py +++ b/text_recognizer/models/transformer.py @@ -13,6 +13,7 @@ from text_recognizer.models.base import BaseLitModel @attr.s(auto_attribs=True) class TransformerLitModel(BaseLitModel): """A PyTorch Lightning model for transformer networks.""" + mapping: Type[AbstractMapping] = attr.ib() start_token: str = attr.ib() end_token: str = attr.ib() |