Update data modules

25 files changed, 263 insertions, 152 deletions

diff --git a/text_recognizer/data/base_dataset.py b/text_recognizer/data/base_dataset.py
index 4ceb818..b840bc8 100644
--- a/text_recognizer/data/base_dataset.py
+++ b/text_recognizer/data/base_dataset.py
@@ -5,8 +5,6 @@ import torch
 from torch import Tensor
 from torch.utils.data import Dataset
 
-from text_recognizer.data.transforms.load_transform import load_transform_from_file
-
 
 class BaseDataset(Dataset):
     r"""Base Dataset class that processes data and targets through optional transfroms.
@@ -23,8 +21,8 @@ class BaseDataset(Dataset):
         self,
         data: Union[Sequence, Tensor],
         targets: Union[Sequence, Tensor],
-        transform: Union[Optional[Callable], str],
-        target_transform: Union[Optional[Callable], str],
+        transform: Callable,
+        target_transform: Callable,
     ) -> None:
         super().__init__()
 
@@ -34,16 +32,6 @@ class BaseDataset(Dataset):
         self.target_transform = target_transform
         if len(self.data) != len(self.targets):
             raise ValueError("Data and targets must be of equal length.")
-        self.transform = self._load_transform(self.transform)
-        self.target_transform = self._load_transform(self.target_transform)
-
-    @staticmethod
-    def _load_transform(
-        transform: Union[Optional[Callable], str]
-    ) -> Optional[Callable]:
-        if isinstance(transform, str):
-            return load_transform_from_file(transform)
-        return transform
 
     def __len__(self) -> int:
         """Return the length of the dataset."""
diff --git a/text_recognizer/data/emnist.py b/text_recognizer/data/emnist.py
index 9c5727f..143705e 100644
--- a/text_recognizer/data/emnist.py
+++ b/text_recognizer/data/emnist.py
@@ -15,7 +15,7 @@ from text_recognizer.data.base_data_module import BaseDataModule, load_and_print
 from text_recognizer.data.base_dataset import BaseDataset, split_dataset
 from text_recognizer.data.transforms.load_transform import load_transform_from_file
 from text_recognizer.data.utils.download_utils import download_dataset
-from text_recognizer.metadata import emnist as metadata
+import text_recognizer.metadata.emnist as metadata
 
 
 class EMNIST(BaseDataModule):
diff --git a/text_recognizer/data/emnist_lines.py b/text_recognizer/data/emnist_lines.py
index 63c9f22..88aac0d 100644
--- a/text_recognizer/data/emnist_lines.py
+++ b/text_recognizer/data/emnist_lines.py
@@ -13,9 +13,9 @@ from text_recognizer.data.base_data_module import BaseDataModule, load_and_print
 from text_recognizer.data.base_dataset import BaseDataset, convert_strings_to_labels
 from text_recognizer.data.emnist import EMNIST
 from text_recognizer.data.mappings import EmnistMapping
-from text_recognizer.data.transforms.load_transform import load_transform_from_file
+from text_recognizer.data.stems.line import LineStem
 from text_recognizer.data.utils.sentence_generator import SentenceGenerator
-from text_recognizer.metadata import emnist_lines as metadata
+import text_recognizer.metadata.emnist_lines as metadata
 
 
 class EMNISTLines(BaseDataModule):
@@ -250,6 +250,6 @@ def _create_dataset_of_images(
 
 def generate_emnist_lines() -> None:
     """Generates a synthetic handwritten dataset and displays info."""
-    transform = load_transform_from_file("transform/emnist_lines.yaml")
-    test_transform = load_transform_from_file("test_transform/default.yaml")
+    transform = LineStem(augment=False)
+    test_transform = LineStem(augment=False)
     load_and_print_info(EMNISTLines(transform=transform, test_transform=test_transform))
diff --git a/text_recognizer/data/iam.py b/text_recognizer/data/iam.py
index c20b50b..2ce1e9c 100644
--- a/text_recognizer/data/iam.py
+++ b/text_recognizer/data/iam.py
@@ -15,7 +15,7 @@ from loguru import logger as log
 
 from text_recognizer.data.base_data_module import BaseDataModule, load_and_print_info
 from text_recognizer.data.utils.download_utils import download_dataset
-from text_recognizer.metadata import iam as metadata
+import text_recognizer.metadata.iam as metadata
 
 
 class IAM(BaseDataModule):
diff --git a/text_recognizer/data/iam_extended_paragraphs.py b/text_recognizer/data/iam_extended_paragraphs.py
index 3ec8221..658626c 100644
--- a/text_recognizer/data/iam_extended_paragraphs.py
+++ b/text_recognizer/data/iam_extended_paragraphs.py
@@ -6,8 +6,10 @@ from torch.utils.data import ConcatDataset
 from text_recognizer.data.base_data_module import BaseDataModule, load_and_print_info
 from text_recognizer.data.iam_paragraphs import IAMParagraphs
 from text_recognizer.data.iam_synthetic_paragraphs import IAMSyntheticParagraphs
+from text_recognizer.data.transforms.pad import Pad
 from text_recognizer.data.mappings import EmnistMapping
-from text_recognizer.data.transforms.load_transform import load_transform_from_file
+from text_recognizer.data.stems.paragraph import ParagraphStem
+import text_recognizer.metadata.iam_paragraphs as metadata
 
 
 class IAMExtendedParagraphs(BaseDataModule):
@@ -104,8 +106,13 @@ class IAMExtendedParagraphs(BaseDataModule):
 
 def show_dataset_info() -> None:
     """Displays Iam extended dataset information."""
-    transform = load_transform_from_file("transform/paragraphs.yaml")
-    test_transform = load_transform_from_file("test_transform/paragraphs_test.yaml")
+    transform = ParagraphStem(augment=False)
+    test_transform = ParagraphStem(augment=False)
+    target_transform = Pad(metadata.MAX_LABEL_LENGTH, 3)
     load_and_print_info(
-        IAMExtendedParagraphs(transform=transform, test_transform=test_transform)
+        IAMExtendedParagraphs(
+            transform=transform,
+            test_transform=test_transform,
+            target_transform=target_transform,
+        )
     )
diff --git a/text_recognizer/data/iam_lines.py b/text_recognizer/data/iam_lines.py
index 3bb189c..e60d1ba 100644
--- a/text_recognizer/data/iam_lines.py
+++ b/text_recognizer/data/iam_lines.py
@@ -21,8 +21,9 @@ from text_recognizer.data.base_dataset import (
 from text_recognizer.data.iam import IAM
 from text_recognizer.data.mappings import EmnistMapping
 from text_recognizer.data.transforms.load_transform import load_transform_from_file
+from text_recognizer.data.stems.line import IamLinesStem
 from text_recognizer.data.utils import image_utils
-from text_recognizer.metadata import iam_lines as metadata
+import text_recognizer.metadata.iam_lines as metadata
 
 ImageFile.LOAD_TRUNCATED_IMAGES = True
 
@@ -227,6 +228,6 @@ def load_line_crops_and_labels(split: str, data_dirname: Path) -> Tuple[List, Li
 
 def generate_iam_lines() -> None:
     """Displays Iam Lines dataset statistics."""
-    transform = load_transform_from_file("transform/lines.yaml")
-    test_transform = load_transform_from_file("test_transform/lines.yaml")
+    transform = IamLinesStem()
+    test_transform = IamLinesStem()
     load_and_print_info(IAMLines(transform=transform, test_transform=test_transform))
diff --git a/text_recognizer/data/iam_paragraphs.py b/text_recognizer/data/iam_paragraphs.py
index eec1b1f..fe1f15c 100644
--- a/text_recognizer/data/iam_paragraphs.py
+++ b/text_recognizer/data/iam_paragraphs.py
@@ -16,9 +16,10 @@ from text_recognizer.data.base_dataset import (
     split_dataset,
 )
 from text_recognizer.data.iam import IAM
+from text_recognizer.data.transforms.pad import Pad
 from text_recognizer.data.mappings import EmnistMapping
-from text_recognizer.data.transforms.load_transform import load_transform_from_file
-from text_recognizer.metadata import iam_paragraphs as metadata
+from text_recognizer.data.stems.paragraph import ParagraphStem
+import text_recognizer.metadata.iam_paragraphs as metadata
 
 
 class IAMParagraphs(BaseDataModule):
@@ -294,8 +295,13 @@ def _num_lines(label: str) -> int:
 
 def create_iam_paragraphs() -> None:
     """Loads and displays dataset statistics."""
-    transform = load_transform_from_file("transform/paragraphs.yaml")
-    test_transform = load_transform_from_file("test_transform/paragraphs_test.yaml")
+    transform = ParagraphStem()
+    test_transform = ParagraphStem()
+    target_transform = Pad(metadata.MAX_LABEL_LENGTH, 3)
     load_and_print_info(
-        IAMParagraphs(transform=transform, test_transform=test_transform)
+        IAMParagraphs(
+            transform=transform,
+            test_transform=test_transform,
+            target_transform=target_transform,
+        )
     )
diff --git a/text_recognizer/data/iam_synthetic_paragraphs.py b/text_recognizer/data/iam_synthetic_paragraphs.py
index 52ed398..91fda4a 100644
--- a/text_recognizer/data/iam_synthetic_paragraphs.py
+++ b/text_recognizer/data/iam_synthetic_paragraphs.py
@@ -9,25 +9,20 @@ from PIL import Image
 from text_recognizer.data.base_data_module import load_and_print_info
 from text_recognizer.data.base_dataset import BaseDataset, convert_strings_to_labels
 from text_recognizer.data.iam import IAM
-from text_recognizer.data.iam_lines import (
-    line_crops_and_labels,
-    load_line_crops_and_labels,
-    save_images_and_labels,
-)
 from text_recognizer.data.iam_paragraphs import (
-    IMAGE_SCALE_FACTOR,
-    NEW_LINE_TOKEN,
     IAMParagraphs,
     get_dataset_properties,
     resize_image,
 )
-from text_recognizer.data.mappings import EmnistMapping
-from text_recognizer.data.transforms.load_transform import load_transform_from_file
-from text_recognizer.metadata import shared as metadata
-
-PROCESSED_DATA_DIRNAME = (
-    metadata.DATA_DIRNAME / "processed" / "iam_synthetic_paragraphs"
+from text_recognizer.data.iam_lines import (
+    line_crops_and_labels,
+    load_line_crops_and_labels,
+    save_images_and_labels,
 )
+from text_recognizer.data.mappings import EmnistMapping
+from text_recognizer.data.stems.paragraph import ParagraphStem
+from text_recognizer.data.transforms.pad import Pad
+import text_recognizer.metadata.iam_synthetic_paragraphs as metadata
 
 
 class IAMSyntheticParagraphs(IAMParagraphs):
@@ -57,26 +52,32 @@ class IAMSyntheticParagraphs(IAMParagraphs):
 
     def prepare_data(self) -> None:
         """Prepare IAM lines to be used to generate paragraphs."""
-        if PROCESSED_DATA_DIRNAME.exists():
+        if metadata.PROCESSED_DATA_DIRNAME.exists():
             return
 
         log.info("Preparing IAM lines for synthetic paragraphs dataset.")
         log.info("Cropping IAM line regions and loading labels.")
 
-        iam = IAM(mapping=EmnistMapping(extra_symbols=(NEW_LINE_TOKEN,)))
+        iam = IAM(mapping=EmnistMapping(extra_symbols=(metadata.NEW_LINE_TOKEN,)))
         iam.prepare_data()
 
         crops_train, labels_train = line_crops_and_labels(iam, "train")
         crops_test, labels_test = line_crops_and_labels(iam, "test")
 
-        crops_train = [resize_image(crop, IMAGE_SCALE_FACTOR) for crop in crops_train]
-        crops_test = [resize_image(crop, IMAGE_SCALE_FACTOR) for crop in crops_test]
+        crops_train = [
+            resize_image(crop, metadata.IMAGE_SCALE_FACTOR) for crop in crops_train
+        ]
+        crops_test = [
+            resize_image(crop, metadata.IMAGE_SCALE_FACTOR) for crop in crops_test
+        ]
 
-        log.info(f"Saving images and labels at {PROCESSED_DATA_DIRNAME}")
+        log.info(f"Saving images and labels at {metadata.PROCESSED_DATA_DIRNAME}")
         save_images_and_labels(
-            crops_train, labels_train, "train", PROCESSED_DATA_DIRNAME
+            crops_train, labels_train, "train", metadata.PROCESSED_DATA_DIRNAME
+        )
+        save_images_and_labels(
+            crops_test, labels_test, "test", metadata.PROCESSED_DATA_DIRNAME
         )
-        save_images_and_labels(crops_test, labels_test, "test", PROCESSED_DATA_DIRNAME)
 
     def setup(self, stage: str = None) -> None:
         """Loading synthetic dataset."""
@@ -85,7 +86,7 @@ class IAMSyntheticParagraphs(IAMParagraphs):
 
         if stage == "fit" or stage is None:
             line_crops, line_labels = load_line_crops_and_labels(
-                "train", PROCESSED_DATA_DIRNAME
+                "train", metadata.PROCESSED_DATA_DIRNAME
             )
             data, paragraphs_labels = generate_synthetic_paragraphs(
                 line_crops=line_crops, line_labels=line_labels
@@ -157,7 +158,7 @@ def generate_synthetic_paragraphs(
 
     paragraphs_crops, paragraphs_labels = [], []
     for paragraph_indices in batched_indices_list:
-        paragraph_label = NEW_LINE_TOKEN.join(
+        paragraph_label = metadata.NEW_LINE_TOKEN.join(
             [line_labels[i] for i in paragraph_indices]
         )
         if len(paragraph_label) > paragraphs_properties["label_length"]["max"]:
@@ -236,8 +237,13 @@ def generate_random_batches(
 
 def create_synthetic_iam_paragraphs() -> None:
     """Creates and prints IAM Synthetic Paragraphs dataset."""
-    transform = load_transform_from_file("transform/paragraphs.yaml")
-    test_transform = load_transform_from_file("test_transform/paragraphs.yaml")
+    transform = ParagraphStem()
+    test_transform = ParagraphStem()
+    target_transform = Pad(metadata.MAX_LABEL_LENGTH, 3)
     load_and_print_info(
-        IAMSyntheticParagraphs(transform=transform, test_transform=test_transform)
+        IAMSyntheticParagraphs(
+            transform=transform,
+            test_transform=test_transform,
+            target_transform=target_transform,
+        )
     )
diff --git a/text_recognizer/data/stems/__init__.py b/text_recognizer/data/stems/__init__.py
new file mode 100644
index 0000000..e69de29
--- /dev/null
+++ b/text_recognizer/data/stems/__init__.py
diff --git a/text_recognizer/data/stems/image.py b/text_recognizer/data/stems/image.py
new file mode 100644
index 0000000..f04b3a0
--- /dev/null
+++ b/text_recognizer/data/stems/image.py
@@ -0,0 +1,18 @@
+from PIL import Image
+import torch
+from torch import Tensor
+import torchvision.transforms as T
+
+
+class ImageStem:
+    def __init__(self) -> None:
+        self.pil_transform = T.Compose([])
+        self.pil_to_tensor = T.ToTensor()
+        self.torch_transform = torch.nn.Sequential()
+
+    def __call__(self, img: Image) -> Tensor:
+        img = self.pil_transform(img)
+        img = self.pil_to_tensor(img)
+        with torch.no_grad():
+            img = self.torch_transform(img)
+        return img
diff --git a/text_recognizer/data/stems/line.py b/text_recognizer/data/stems/line.py
new file mode 100644
index 0000000..2fe1a2c
--- /dev/null
+++ b/text_recognizer/data/stems/line.py
@@ -0,0 +1,86 @@
+import random
+
+from PIL import Image
+import torchvision.transforms as T
+
+import text_recognizer.metadata.iam_lines as metadata
+from text_recognizer.data.stems.image import ImageStem
+
+
+class LineStem(ImageStem):
+    """A stem for handling images containing a line of text."""
+
+    def __init__(
+        self, augment=False, color_jitter_kwargs=None, random_affine_kwargs=None
+    ):
+        super().__init__()
+        if color_jitter_kwargs is None:
+            color_jitter_kwargs = {"brightness": (0.5, 1)}
+        if random_affine_kwargs is None:
+            random_affine_kwargs = {
+                "degrees": 3,
+                "translate": (0, 0.05),
+                "scale": (0.4, 1.1),
+                "shear": (-40, 50),
+                "interpolation": T.InterpolationMode.BILINEAR,
+                "fill": 0,
+            }
+
+        if augment:
+            self.pil_transforms = T.Compose(
+                [
+                    T.ColorJitter(**color_jitter_kwargs),
+                    T.RandomAffine(**random_affine_kwargs),
+                ]
+            )
+
+
+class IamLinesStem(ImageStem):
+    """A stem for handling images containing lines of text from the IAMLines dataset."""
+
+    def __init__(
+        self, augment=False, color_jitter_kwargs=None, random_affine_kwargs=None
+    ):
+        super().__init__()
+
+        def embed_crop(crop, augment=augment):
+            # crop is PIL.image of dtype="L" (so values range from 0 -> 255)
+            image = Image.new("L", (metadata.IMAGE_WIDTH, metadata.IMAGE_HEIGHT))
+
+            # Resize crop
+            crop_width, crop_height = crop.size
+            new_crop_height = metadata.IMAGE_HEIGHT
+            new_crop_width = int(new_crop_height * (crop_width / crop_height))
+            if augment:
+                # Add random stretching
+                new_crop_width = int(new_crop_width * random.uniform(0.9, 1.1))
+                new_crop_width = min(new_crop_width, metadata.IMAGE_WIDTH)
+            crop_resized = crop.resize(
+                (new_crop_width, new_crop_height), resample=Image.BILINEAR
+            )
+
+            # Embed in the image
+            x = min(metadata.CHAR_WIDTH, metadata.IMAGE_WIDTH - new_crop_width)
+            y = metadata.IMAGE_HEIGHT - new_crop_height
+
+            image.paste(crop_resized, (x, y))
+
+            return image
+
+        if color_jitter_kwargs is None:
+            color_jitter_kwargs = {"brightness": (0.8, 1.6)}
+        if random_affine_kwargs is None:
+            random_affine_kwargs = {
+                "degrees": 1,
+                "shear": (-30, 20),
+                "interpolation": T.InterpolationMode.BILINEAR,
+                "fill": 0,
+            }
+
+        pil_transform_list = [T.Lambda(embed_crop)]
+        if augment:
+            pil_transform_list += [
+                T.ColorJitter(**color_jitter_kwargs),
+                T.RandomAffine(**random_affine_kwargs),
+            ]
+        self.pil_transform = T.Compose(pil_transform_list)
diff --git a/text_recognizer/data/stems/paragraph.py b/text_recognizer/data/stems/paragraph.py
new file mode 100644
index 0000000..39e1e59
--- /dev/null
+++ b/text_recognizer/data/stems/paragraph.py
@@ -0,0 +1,66 @@
+"""Iam paragraph stem class."""
+import torchvision.transforms as T
+
+import text_recognizer.metadata.iam_paragraphs as metadata
+from text_recognizer.data.stems.image import ImageStem
+
+
+IMAGE_HEIGHT, IMAGE_WIDTH = metadata.IMAGE_HEIGHT, metadata.IMAGE_WIDTH
+IMAGE_SHAPE = metadata.IMAGE_SHAPE
+
+MAX_LABEL_LENGTH = metadata.MAX_LABEL_LENGTH
+
+
+class ParagraphStem(ImageStem):
+    """A stem for handling images that contain a paragraph of text."""
+
+    def __init__(
+        self,
+        augment=False,
+        color_jitter_kwargs=None,
+        random_affine_kwargs=None,
+        random_perspective_kwargs=None,
+        gaussian_blur_kwargs=None,
+        sharpness_kwargs=None,
+    ):
+        super().__init__()
+
+        if not augment:
+            self.pil_transform = T.Compose([T.CenterCrop(IMAGE_SHAPE)])
+        else:
+            if color_jitter_kwargs is None:
+                color_jitter_kwargs = {"brightness": 0.4, "contrast": 0.4}
+            if random_affine_kwargs is None:
+                random_affine_kwargs = {
+                    "degrees": 3,
+                    "shear": 6,
+                    "scale": (0.95, 1),
+                    "interpolation": T.InterpolationMode.BILINEAR,
+                }
+            if random_perspective_kwargs is None:
+                random_perspective_kwargs = {
+                    "distortion_scale": 0.2,
+                    "p": 0.5,
+                    "interpolation": T.InterpolationMode.BILINEAR,
+                }
+            if gaussian_blur_kwargs is None:
+                gaussian_blur_kwargs = {"kernel_size": (3, 3), "sigma": (0.1, 1.0)}
+            if sharpness_kwargs is None:
+                sharpness_kwargs = {"sharpness_factor": 2, "p": 0.5}
+
+            self.pil_transform = T.Compose(
+                [
+                    T.ColorJitter(**color_jitter_kwargs),
+                    T.RandomCrop(
+                        size=IMAGE_SHAPE,
+                        padding=None,
+                        pad_if_needed=True,
+                        fill=0,
+                        padding_mode="constant",
+                    ),
+                    T.RandomAffine(**random_affine_kwargs),
+                    T.RandomPerspective(**random_perspective_kwargs),
+                    T.GaussianBlur(**gaussian_blur_kwargs),
+                    T.RandomAdjustSharpness(**sharpness_kwargs),
+                ]
+            )
diff --git a/text_recognizer/metadata/iam_lines.py b/text_recognizer/metadata/iam_lines.py
index 73b2a28..a94041c 100644
--- a/text_recognizer/metadata/iam_lines.py
+++ b/text_recognizer/metadata/iam_lines.py
@@ -9,7 +9,12 @@ CHAR_WIDTH = emnist.INPUT_SHAPE[0] // IMAGE_SCALE_FACTOR  # rough estimate
 IMAGE_HEIGHT = 112 // IMAGE_SCALE_FACTOR
 IMAGE_WIDTH = 3072 // IMAGE_SCALE_FACTOR  # rounding up IAMLines empirical maximum width
 
+SEED = 4711
 DIMS = (1, IMAGE_HEIGHT, IMAGE_WIDTH)
 OUTPUT_DIMS = (89, 1)
+IMAGE_HEIGHT = 56
+IMAGE_WIDTH = 1024
+MAX_LABEL_LENGTH = 89
+MAX_WORD_PIECE_LENGTH = 72
 
 MAPPING = emnist.MAPPING
diff --git a/text_recognizer/metadata/iam_paragraphs.py b/text_recognizer/metadata/iam_paragraphs.py
index 70b802b..7bb909a 100644
--- a/text_recognizer/metadata/iam_paragraphs.py
+++ b/text_recognizer/metadata/iam_paragraphs.py
@@ -8,11 +8,13 @@ NEW_LINE_TOKEN = "\n"
 MAPPING = [*emnist.MAPPING, NEW_LINE_TOKEN]
 
 # must match IMAGE_SCALE_FACTOR for IAMLines to be compatible with synthetic paragraphs
+SEED = 4711
 IMAGE_SCALE_FACTOR = 2
+IMAGE_HEIGHT = 1152 // IMAGE_SCALE_FACTOR
+IMAGE_WIDTH = 1280 // IMAGE_SCALE_FACTOR
+MAX_LABEL_LENGTH = 682
 IMAGE_HEIGHT, IMAGE_WIDTH = 576, 640
 IMAGE_SHAPE = (IMAGE_HEIGHT, IMAGE_WIDTH)
 
-MAX_LABEL_LENGTH = 682
-
 DIMS = (1, IMAGE_HEIGHT, IMAGE_WIDTH)
 OUTPUT_DIMS = (MAX_LABEL_LENGTH, 1)
diff --git a/text_recognizer/metadata/iam_synthetic_paragraphs.py b/text_recognizer/metadata/iam_synthetic_paragraphs.py
index 15bcfc8..6c11d28 100644
--- a/text_recognizer/metadata/iam_synthetic_paragraphs.py
+++ b/text_recognizer/metadata/iam_synthetic_paragraphs.py
@@ -2,5 +2,7 @@ import text_recognizer.metadata.iam_paragraphs as iam_paragraphs
 import text_recognizer.metadata.shared as shared
 
 NEW_LINE_TOKEN = iam_paragraphs.NEW_LINE_TOKEN
-
+DATA_DIRNAME = shared.DATA_DIRNAME
 PROCESSED_DATA_DIRNAME = shared.DATA_DIRNAME / "processed" / "iam_synthetic_paragraphs"
+
+MAX_LABEL_LENGTH = 682
diff --git a/training/conf/datamodule/iam_extended_paragraphs.yaml b/training/conf/datamodule/iam_extended_paragraphs.yaml
index 0921b25..c46714c 100644
--- a/training/conf/datamodule/iam_extended_paragraphs.yaml
+++ b/training/conf/datamodule/iam_extended_paragraphs.yaml
@@ -3,9 +3,16 @@ batch_size: 4
 num_workers: 12
 train_fraction: 0.8
 pin_memory: true
-transform: transform/paragraphs.yaml
-test_transform: test_transform/paragraphs.yaml
-target_transform: target_transform/pad.yaml
+transform:
+  _target_: text_recognizer.data.stems.paragraph.ParagraphStem
+  augment: true
+test_transform:
+  _target_: text_recognizer.data.stems.paragraph.ParagraphStem
+  augment: false
+target_transform:
+  _target_: text_recognizer.data.transforms.pad.Pad
+  max_len: 682
+  pad_index: 3
 mapping:
   _target_: text_recognizer.data.mappings.EmnistMapping
   extra_symbols: ["\n"]
diff --git a/training/conf/datamodule/iam_lines.yaml b/training/conf/datamodule/iam_lines.yaml
index 85a023b..4f1f1b8 100644
--- a/training/conf/datamodule/iam_lines.yaml
+++ b/training/conf/datamodule/iam_lines.yaml
@@ -3,7 +3,11 @@ batch_size: 8
 num_workers: 12
 train_fraction: 0.9
 pin_memory: true
-transform: transform/lines.yaml
-test_transform: test_transform/lines.yaml
+transform:
+  _target_: text_recognizer.data.stems.line.IamLinesStem
+  augment: true
+test_transform:
+  _target_: text_recognizer.data.stems.line.IamLinesStem
+  augment: false
 mapping:
   _target_: text_recognizer.data.mappings.EmnistMapping
diff --git a/training/conf/datamodule/target_transform/pad.yaml b/training/conf/datamodule/target_transform/pad.yaml
deleted file mode 100644
index 854c24c..0000000
--- a/training/conf/datamodule/target_transform/pad.yaml
+++ /dev/null
@@ -1,4 +0,0 @@
-pad:
-  _target_: text_recognizer.data.transforms.pad.Pad
-  max_len: 682
-  pad_index: 3
diff --git a/training/conf/datamodule/test_transform/default.yaml b/training/conf/datamodule/test_transform/default.yaml
deleted file mode 100644
index d6f6d78..0000000
--- a/training/conf/datamodule/test_transform/default.yaml
+++ /dev/null
@@ -1,2 +0,0 @@
-to_tensor:
-  _target_: torchvision.transforms.ToTensor
diff --git a/training/conf/datamodule/test_transform/lines.yaml b/training/conf/datamodule/test_transform/lines.yaml
deleted file mode 100644
index a14ec82..0000000
--- a/training/conf/datamodule/test_transform/lines.yaml
+++ /dev/null
@@ -1,6 +0,0 @@
-embed_crop:
-  _target_: text_recognizer.data.transforms.embed_crop.EmbedCrop
-  augment: true
-
-to_tensor:
-  _target_: torchvision.transforms.ToTensor
diff --git a/training/conf/datamodule/test_transform/paragraphs.yaml b/training/conf/datamodule/test_transform/paragraphs.yaml
deleted file mode 100644
index a69eae9..0000000
--- a/training/conf/datamodule/test_transform/paragraphs.yaml
+++ /dev/null
@@ -1,6 +0,0 @@
-center_crop:
-  _target_: torchvision.transforms.CenterCrop
-  size: [576, 640]
-
-to_tensor:
-  _target_: torchvision.transforms.ToTensor
diff --git a/training/conf/datamodule/transform/default.yaml b/training/conf/datamodule/transform/default.yaml
deleted file mode 100644
index d6f6d78..0000000
--- a/training/conf/datamodule/transform/default.yaml
+++ /dev/null
@@ -1,2 +0,0 @@
-to_tensor:
-  _target_: torchvision.transforms.ToTensor
diff --git a/training/conf/datamodule/transform/emnist_lines.yaml b/training/conf/datamodule/transform/emnist_lines.yaml
deleted file mode 100644
index 986ba20..0000000
--- a/training/conf/datamodule/transform/emnist_lines.yaml
+++ /dev/null
@@ -1,14 +0,0 @@
-to_tensor:
-  _target_: torchvision.transforms.ToTensor
-
-color_jitter:
-  _target_: torchvision.transforms.ColorJitter
-  brightness: [0.5, 1.0]
-
-random_affine:
-  _target_: torchvision.transforms.RandomAffine
-  degrees: 3
-  translate: [0.0, 0.05]
-  scale: [0.4, 1.1]
-  shear: [-40, 50]
-  fill: 0
diff --git a/training/conf/datamodule/transform/lines.yaml b/training/conf/datamodule/transform/lines.yaml
deleted file mode 100644
index 6bd35ef..0000000
--- a/training/conf/datamodule/transform/lines.yaml
+++ /dev/null
@@ -1,23 +0,0 @@
-embed_crop:
-  _target_: text_recognizer.data.transforms.embed_crop.EmbedCrop
-  augment: false
-
-color_jitter:
-  _target_: torchvision.transforms.ColorJitter
-  brightness: [0.8, 1.6]
-
-random_affine:
-  _target_: torchvision.transforms.RandomAffine
-  degrees: 1
-  shear: [-10, 10]
-  interpolation: BILINEAR
-  fill: 0
-
-random_perspective:
-  _target_: torchvision.transforms.RandomPerspective
-  distortion_scale: 0.05
-  p: 0.25
-  fill: 0
-
-to_tensor:
-  _target_: torchvision.transforms.ToTensor
diff --git a/training/conf/datamodule/transform/paragraphs.yaml b/training/conf/datamodule/transform/paragraphs.yaml
deleted file mode 100644
index 4f260d0..0000000
--- a/training/conf/datamodule/transform/paragraphs.yaml
+++ /dev/null
@@ -1,30 +0,0 @@
-# random_crop:
-#   _target_: torchvision.transforms.RandomCrop
-#   size: [576, 640]
-#   padding: null
-#   pad_if_needed: true
-#   fill: 0
-#   padding_mode: constant
-
-center_crop:
-  _target_: torchvision.transforms.CenterCrop
-  size: [576, 640]
-
-color_jitter:
-  _target_: torchvision.transforms.ColorJitter
-  brightness: [0.8, 1.6]
-
-random_affine:
-  _target_: torchvision.transforms.RandomAffine
-  degrees: 1
-  shear: [-5, 5]
-  interpolation: BILINEAR
-
-random_perspective:
-  _target_: torchvision.transforms.RandomPerspective
-  distortion_scale: 0.05
-  p: 0.25
-  fill: 0
-
-to_tensor:
-  _target_: torchvision.transforms.ToTensor