dan/manager/ocr.py

@@ -9,7 +9,6 @@ import torch
 from fontTools.ttLib import TTFont
 from PIL import Image, ImageDraw, ImageFont
 
-from dan import logger
 from dan.manager.dataset import DatasetManager, GenericDataset, apply_preprocessing
 from dan.ocr.utils import LM_str_to_ind
 from dan.utils import (
@@ -42,8 +41,6 @@ class OCRDatasetManager(DatasetManager):
             and self.params["config"]["synthetic_data"]
         ):
             self.synthetic_data = self.params["config"]["synthetic_data"]
-            if "config" in self.synthetic_data:
-                self.synthetic_data["config"]["valid_fonts"] = self.get_valid_fonts()
 
         if "new_tokens" in params:
             self.charset = sorted(
@@ -104,34 +101,6 @@ class OCRDatasetManager(DatasetManager):
                 [s["img"].shape[1] for s in self.train_dataset.samples]
             )
 
-    def get_valid_fonts(self):
-        """
-        Select fonts that are compatible with the alphabet
-        """
-        font_path = self.synthetic_data["font_path"]
-        alphabet = self.charset.copy()
-        special_chars = ["\n"]
-        alphabet = [char for char in alphabet if char not in special_chars]
-        valid_fonts = list()
-        for fold_detail in os.walk(font_path):
-            if fold_detail[2]:
-                for font_name in fold_detail[2]:
-                    if ".ttf" not in font_name:
-                        continue
-                    font_path = os.path.join(fold_detail[0], font_name)
-                    to_add = True
-                    if alphabet is not None:
-                        for char in alphabet:
-                            if not char_in_font(char, font_path):
-                                to_add = False
-                                break
-                        if to_add:
-                            valid_fonts.append(font_path)
-                    else:
-                        valid_fonts.append(font_path)
-        logger.info(f"Found {len(valid_fonts)} fonts.")
-        return valid_fonts
-
 
 class OCRDataset(GenericDataset):
     """

dan/manager/training.py

@@ -1042,7 +1042,6 @@ class OCRManager(GenericTrainingManager):
                                 {
                                     "path": sample["path"],
                                     "label": chunk,
-                                    "nb_cols": 1,
                                 }
                             )
 
@@ -1055,7 +1054,6 @@ class OCRManager(GenericTrainingManager):
                 Image.fromarray(img).save(img_path)
                 gt[set_name][img_name] = {
                     "text": sample["label"],
-                    "nb_cols": sample["nb_cols"] if "nb_cols" in sample else 1,
                 }
                 if "line_label" in sample:
                     gt[set_name][img_name]["lines"] = sample["line_label"]