Add weighted type sampling reflecting Korean road LP distribution

Default weights (Type 1/2/3/4 = 0.80/0.05/0.10/0.05) approximate the
real-world mix: ~98% of registered cars are post-2006 single-line, plus
~7.5% commercial yellow. Override via --type_weights when needed. Also
prints per-type counts so the operator can sanity-check distribution.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

data_gen/generate_synthetic.py

@@ -175,6 +175,11 @@ def make_label_two_line(plate, top, bot):
     ]
 
 
+# 한국 도로 LP 분포 추정 (자가용 92% + 영업용 7.5%, 신형 가로 ~98% 등)
+# 자산 한계 고려한 합성 가중치 — generate_synthetic 호출 시 --type_weights 로 덮어쓰기 가능.
+TYPE_DEFAULT_WEIGHTS = {'1': 0.80, '2': 0.05, '3': 0.10, '4': 0.05}
+
+
 def main():
     p = argparse.ArgumentParser(description=__doc__)
     p.add_argument("--asset_dir", default=str(Path(__file__).parent / "Korean-license-plate-Generator"))
@@ -182,6 +187,8 @@ def main():
     p.add_argument("--num", type=int, default=200, help="총 이미지 개수")
     p.add_argument("--test_ratio", type=float, default=0.05)
     p.add_argument("--types", default="1,2,3,4", help="포함할 type (콤마 구분)")
+    p.add_argument("--type_weights", default=None,
+                   help="--types 순서대로 콤마 구분 가중치. 미지정 시 한국 도로 분포 기본값 사용.")
     p.add_argument("--dict", default=None, help="검증용 dict 경로 (선택)")
     p.add_argument("--seed", type=int, default=42)
     args = p.parse_args()
@@ -196,10 +203,22 @@ def main():
         '3': ('two', gen.gen_type3),
         '4': ('two', gen.gen_type4),
     }
-    chosen = [type_funcs[t.strip()] for t in args.types.split(',') if t.strip() in type_funcs]
-    if not chosen:
+    selected_keys = [t.strip() for t in args.types.split(',') if t.strip() in type_funcs]
+    if not selected_keys:
         sys.exit("No valid types selected.")
-
+    chosen = [type_funcs[k] for k in selected_keys]
+
+    if args.type_weights:
+        weights = [float(w) for w in args.type_weights.split(',')]
+        if len(weights) != len(selected_keys):
+            sys.exit(f"--type_weights ({len(weights)}) length mismatch with --types ({len(selected_keys)})")
+    else:
+        weights = [TYPE_DEFAULT_WEIGHTS[k] for k in selected_keys]
+    total = sum(weights)
+    weights = [w / total for w in weights]
+    print(f"Type sampling weights: {dict(zip(selected_keys, [round(w, 3) for w in weights]))}")
+
+    type_count = {k: 0 for k in selected_keys}
     seen_chars = set()
     out = Path(args.out_dir)
     n_test = max(1, int(args.num * args.test_ratio))
@@ -210,7 +229,9 @@ def main():
         img_dir.mkdir(parents=True, exist_ok=True)
         records = []
         for i in range(count):
-            kind, fn = random.choice(chosen)
+            idx = random.choices(range(len(chosen)), weights=weights, k=1)[0]
+            kind, fn = chosen[idx]
+            type_count[selected_keys[idx]] += 1
             if kind == 'one':
                 plate, text = fn()
                 label = make_label_one_line(plate, text)
@@ -228,6 +249,8 @@ def main():
                 f.write(f"{path}\t{lab}\n")
         print(f"  {split}: {len(records)} images → {out / split}")
 
+    print(f"Type counts: {type_count}")
+
     # dict 검증
     if args.dict:
         with open(args.dict, encoding="utf-8") as f: