feat: add exclusion rules

configs/font.yml

@@ -11,6 +11,9 @@
 # - Others
 dataset:
   path: ./dataset/fonts
+  exclusion:
+    - ./Founder Type（方正）/韩文/方正朝文中圆.TTF
+    - ./Founder Type（方正）/简繁/ttf/方正宋体S-超大字符集(SIP).TTF
   specs:
     - path:
         - ./Adobe/CJK

font_dataset/font.py

@@ -1,6 +1,5 @@
 import yaml
 import os
-from typing import List
 
 
 from .utils import get_files
@@ -15,7 +14,7 @@ class DSFont:
         self.language = language
 
 
-def load_fonts(config_path="configs/font.yml") -> List[DSFont]:
+def load_fonts(config_path="configs/font.yml"):
     with open(config_path, "r", encoding="utf-8") as f:
         config = yaml.safe_load(f)
 
@@ -41,4 +40,14 @@ def load_fonts(config_path="configs/font.yml") -> List[DSFont]:
                 font_list.append(DSFont(file, spec["language"]))
 
     font_list.sort(key=lambda x: x.path)
-    return font_list
+
+    exclusion_list = ds_config["exclusion"]
+    exclusion_list = [os.path.join(ds_path, path) for path in exclusion_list]
+
+    def exclusion_rule(font: DSFont):
+        for exclusion in exclusion_list:
+            if os.path.samefile(font.path, exclusion):
+                return True
+        return False
+
+    return font_list, exclusion_rule

font_ds_generate_script.py

@@ -31,7 +31,7 @@ dataset_path = "./dataset/font_img"
 os.makedirs(dataset_path, exist_ok=True)
 
 
-fonts = load_fonts()
+fonts, exclusion_rule = load_fonts()
 corpus_manager = CorpusGeneratorManager()
 images = background_image_generator()
 
@@ -41,10 +41,17 @@ def generate_dataset(dataset_type: str, cnt: int):
     os.makedirs(dataset_bath_dir, exist_ok=True)
 
     def _generate_single(args):
+        i, j, font = args
+        print(
+            f"Generating {dataset_type} font: {font.path} {i} / {len(fonts)}, image {j}"
+        )
+
+        if exclusion_rule(font):
+            print(f"Excluded font: {font.path}")
+            return
+
         while True:
             try:
-                i, j, font = args
-
                 image_file_name = f"font_{i}_img_{j}.jpg"
                 label_file_name = f"font_{i}_img_{j}.bin"
 

font_ds_stat.py

@@ -23,12 +23,16 @@ test_cnt_cjk = int(test_cnt * cjk_ratio)
 dataset_path = "./dataset/font_img"
 os.makedirs(dataset_path, exist_ok=True)
 
-fonts = load_fonts()
+fonts, exclusion_rule = load_fonts()
 
 
 cnt = 0
 
 for font in fonts:
+    if exclusion_rule(font):
+        print(f"Excluded font: {font.path}")
+        continue
+
     if font.language == "CJK":
         cnt += cjk_ratio
     else: