feat : 데이터 분할, 로그 기록 추가

.claude/settings.local.json

@@ -0,0 +1,17 @@
+{
+  "permissions": {
+    "allow": [
+      "Bash(python split_dataset_fixed.py:*)",
+      "Bash(python -c:*)",
+      "Bash(tee:*)",
+      "Bash(python split_simple.py:*)",
+      "Bash(python split_final.py:*)",
+      "Bash(dir:*)",
+      "Bash(powershell \"(Get-ChildItem dataset/images/train).Count\")",
+      "Bash(python split_correct.py:*)",
+      "Bash(powershell \"(Get-ChildItem dataset/images/train).Count; (Get-ChildItem dataset/images/val).Count; (Get-ChildItem dataset/images/test).Count\")"
+    ],
+    "deny": [],
+    "ask": []
+  }
+}

split_correct.py

@@ -0,0 +1,117 @@
+"""
+데이터셋 분할 - Windows 대소문자 문제 해결
+각 파일에 고유한 번호를 부여하여 충돌 방지
+"""
+import os
+import shutil
+from pathlib import Path
+import random
+
+random.seed(42)
+
+# 1. 모든 이미지-라벨 쌍 찾기
+print("Finding all image-label pairs...")
+pairs = []
+
+for batch_num in range(1, 18):
+    batch_name = f'batch_{batch_num}'
+    img_dir = Path(f'data/{batch_name}')
+    lbl_dir = Path(f'labels/{batch_name}')
+
+    if not img_dir.exists():
+        continue
+
+    # 라벨 파일 기준
+    for lbl_file in lbl_dir.glob('*.txt'):
+        if lbl_file.name == 'classes.txt':
+            continue
+
+        stem = lbl_file.stem
+
+        # 이미지 파일 찾기
+        img_file_jpg = img_dir / f'{stem}.jpg'
+        img_file_JPG = img_dir / f'{stem}.JPG'
+
+        img_file = None
+        if img_file_jpg.exists():
+            img_file = img_file_jpg
+        elif img_file_JPG.exists():
+            img_file = img_file_JPG
+
+        if img_file and img_file.exists():
+            pairs.append({
+                'image': str(img_file),
+                'label': str(lbl_file),
+                'stem': stem,
+                'ext': img_file.suffix
+            })
+
+print(f"Found {len(pairs)} valid pairs")
+
+# 2. 분할
+random.shuffle(pairs)
+
+total = len(pairs)
+train_size = int(total * 0.8)
+val_size = int(total * 0.1)
+
+train_pairs = pairs[:train_size]
+val_pairs = pairs[train_size:train_size + val_size]
+test_pairs = pairs[train_size + val_size:]
+
+print(f"Train: {len(train_pairs)}, Val: {len(val_pairs)}, Test: {len(test_pairs)}")
+
+# 3. 디렉토리 생성
+os.makedirs('dataset/images/train', exist_ok=True)
+os.makedirs('dataset/images/val', exist_ok=True)
+os.makedirs('dataset/images/test', exist_ok=True)
+os.makedirs('dataset/labels/train', exist_ok=True)
+os.makedirs('dataset/labels/val', exist_ok=True)
+os.makedirs('dataset/labels/test', exist_ok=True)
+
+# 4. 파일 복사 - 고유한 번호 부여
+def copy_files(pairs, split):
+    print(f"\nCopying {split}...")
+    for idx, pair in enumerate(pairs):
+        if (idx + 1) % 200 == 0:
+            print(f"  {idx+1}/{len(pairs)}...")
+
+        # 고유한 파일명 생성 (idx를 이용)
+        # {idx:05d}_{stem}{ext} 형식
+        img_dst = f"dataset/images/{split}/{idx:05d}_{pair['stem']}{pair['ext']}"
+        lbl_dst = f"dataset/labels/{split}/{idx:05d}_{pair['stem']}.txt"
+
+        shutil.copy2(pair['image'], img_dst)
+        shutil.copy2(pair['label'], lbl_dst)
+
+    print(f"  Done! {len(pairs)} files copied")
+
+    # 검증
+    actual_count = len(os.listdir(f"dataset/images/{split}"))
+    print(f"  Verified: {actual_count} files in directory")
+
+    if actual_count != len(pairs):
+        print(f"  WARNING: Expected {len(pairs)} but found {actual_count}")
+
+copy_files(train_pairs, 'train')
+copy_files(val_pairs, 'val')
+copy_files(test_pairs, 'test')
+
+# 5. data.yaml
+with open('dataset/data.yaml', 'w') as f:
+    f.write(f"""# TACO Waste Classification Dataset
+path: {Path('dataset').absolute()}
+train: images/train
+val: images/val
+test: images/test
+
+nc: 5
+names: ['Plastic', 'Vinyl', 'Can', 'Glass', 'Paper']
+""")
+
+print("\ndata.yaml created")
+print("\n=== DONE ===")
+print(f"\nFinal verification:")
+print(f"  Train: {len(os.listdir('dataset/images/train'))} images, {len(os.listdir('dataset/labels/train'))} labels")
+print(f"  Val: {len(os.listdir('dataset/images/val'))} images, {len(os.listdir('dataset/labels/val'))} labels")
+print(f"  Test: {len(os.listdir('dataset/images/test'))} images, {len(os.listdir('dataset/labels/test'))} labels")

split_log.txt

@@ -0,0 +1,35 @@
+Finding all image-label pairs...
+Found 1500 unique image-label pairs
+
+Splitting dataset (80% train, 10% val, 10% test)...
+  Train: 1200 pairs (should be 1200)
+  Val:   150 pairs (should be 150)
+  Test:  150 pairs (should be 150)
+
+Copying 1200 files to train...
+
  train:   0%|          | 0/1200 [00:00<?, ?it/s]
  train:   1%|1         | 15/1200 [00:00<00:08, 147.73it/s]
  train:   3%|2         | 31/1200 [00:00<00:07, 149.66it/s]
  train:   4%|3         | 46/1200 [00:00<00:07, 148.04it/s]
  train:   5%|5         | 61/1200 [00:00<00:07, 145.73it/s]
  train:   6%|6         | 76/1200 [00:00<00:07, 144.29it/s]
  train:   8%|7         | 91/1200 [00:00<00:07, 144.02it/s]
  train:   9%|8         | 106/1200 [00:00<00:07, 143.16it/s]
  train:  10%|#         | 121/1200 [00:00<00:07, 143.70it/s]
  train:  11%|#1        | 136/1200 [00:00<00:07, 138.66it/s]
  train:  12%|#2        | 150/1200 [00:01<00:07, 137.23it/s]
  train:  14%|#3        | 164/1200 [00:01<00:07, 135.00it/s]
  train:  15%|#5        | 180/1200 [00:01<00:07, 139.19it/s]
  train:  16%|#6        | 194/1200 [00:01<00:07, 137.42it/s]
  train:  17%|#7        | 208/1200 [00:01<00:07, 136.94it/s]
  train:  18%|#8        | 222/1200 [00:01<00:07, 135.75it/s]
  train:  20%|#9        | 237/1200 [00:01<00:06, 139.65it/s]
  train:  21%|##        | 251/1200 [00:01<00:07, 129.33it/s]
  train:  22%|##2       | 265/1200 [00:01<00:07, 130.33it/s]
  train:  23%|##3       | 280/1200 [00:02<00:06, 134.53it/s]
  train:  25%|##4       | 296/1200 [00:02<00:06, 140.81it/s]
  train:  26%|##6       | 312/1200 [00:02<00:06, 144.65it/s]
  train:  27%|##7       | 327/1200 [00:02<00:06, 138.37it/s]
  train:  28%|##8       | 341/1200 [00:02<00:06, 135.72it/s]
  train:  30%|##9       | 355/1200 [00:02<00:06, 129.09it/s]
  train:  31%|###       | 370/1200 [00:02<00:06, 131.39it/s]
  train:  32%|###2      | 385/1200 [00:02<00:06, 134.27it/s]
  train:  33%|###3      | 400/1200 [00:02<00:05, 137.87it/s]
  train:  34%|###4      | 414/1200 [00:03<00:05, 131.67it/s]
  train:  36%|###5      | 429/1200 [00:03<00:05, 134.05it/s]
  train:  37%|###7      | 444/1200 [00:03<00:05, 136.15it/s]
  train:  38%|###8      | 459/1200 [00:03<00:05, 138.04it/s]
  train:  39%|###9      | 473/1200 [00:03<00:05, 138.46it/s]
  train:  41%|####      | 487/1200 [00:03<00:05, 138.48it/s]
  train:  42%|####1     | 501/1200 [00:03<00:05, 137.95it/s]
  train:  43%|####2     | 515/1200 [00:03<00:05, 135.56it/s]
  train:  44%|####4     | 530/1200 [00:03<00:04, 137.32it/s]
  train:  45%|####5     | 544/1200 [00:03<00:04, 135.31it/s]
  train:  46%|####6     | 558/1200 [00:04<00:04, 133.81it/s]
  train:  48%|####7     | 573/1200 [00:04<00:04, 136.71it/s]
  train:  49%|####8     | 587/1200 [00:04<00:04, 137.45it/s]
  train:  50%|#####     | 601/1200 [00:04<00:04, 137.15it/s]
  train:  51%|#####1    | 616/1200 [00:04<00:04, 139.47it/s]
  train:  52%|#####2    | 630/1200 [00:04<00:04, 137.44it/s]
  train:  54%|#####3    | 645/1200 [00:04<00:03, 140.06it/s]
  train:  55%|#####5    | 660/1200 [00:04<00:03, 139.09it/s]
  train:  56%|#####6    | 674/1200 [00:04<00:03, 133.37it/s]
  train:  57%|#####7    | 688/1200 [00:05<00:04, 127.29it/s]
  train:  58%|#####8    | 701/1200 [00:05<00:03, 127.51it/s]
  train:  60%|#####9    | 715/1200 [00:05<00:03, 129.37it/s]
  train:  61%|######    | 728/1200 [00:05<00:03, 127.30it/s]
  train:  62%|######2   | 744/1200 [00:05<00:03, 136.08it/s]
  train:  63%|######3   | 759/1200 [00:05<00:03, 139.08it/s]
  train:  64%|######4   | 773/1200 [00:05<00:03, 138.73it/s]
  train:  66%|######5   | 787/1200 [00:05<00:03, 133.05it/s]
  train:  67%|######6   | 801/1200 [00:05<00:03, 131.26it/s]
  train:  68%|######7   | 815/1200 [00:05<00:02, 129.68it/s]
  train:  69%|######9   | 830/1200 [00:06<00:02, 135.20it/s]
  train:  70%|#######   | 844/1200 [00:06<00:02, 128.28it/s]
  train:  72%|#######1  | 858/1200 [00:06<00:02, 131.50it/s]
  train:  73%|#######2  | 872/1200 [00:06<00:02, 132.38it/s]
  train:  74%|#######3  | 886/1200 [00:06<00:02, 134.00it/s]
  train:  75%|#######5  | 900/1200 [00:06<00:02, 126.39it/s]
  train:  76%|#######6  | 914/1200 [00:06<00:02, 129.61it/s]
  train:  77%|#######7  | 928/1200 [00:06<00:02, 128.12it/s]
  train:  78%|#######8  | 941/1200 [00:06<00:02, 128.59it/s]
  train:  80%|#######9  | 954/1200 [00:07<00:01, 126.27it/s]
  train:  81%|########  | 967/1200 [00:07<00:01, 124.99it/s]
  train:  82%|########1 | 983/1200 [00:07<00:01, 132.53it/s]
  train:  83%|########3 | 997/1200 [00:07<00:01, 133.82it/s]
  train:  84%|########4 | 1012/1200 [00:07<00:01, 137.23it/s]
  train:  86%|########5 | 1028/1200 [00:07<00:01, 140.46it/s]
  train:  87%|########6 | 1043/1200 [00:07<00:01, 138.32it/s]
  train:  88%|########8 | 1057/1200 [00:07<00:01, 135.13it/s]
  train:  89%|########9 | 1071/1200 [00:07<00:00, 132.91it/s]
  train:  90%|######### | 1085/1200 [00:08<00:00, 126.40it/s]
  train:  92%|#########1| 1100/1200 [00:08<00:00, 130.89it/s]
  train:  93%|#########2| 1114/1200 [00:08<00:00, 126.20it/s]
  train:  94%|#########3| 1128/1200 [00:08<00:00, 129.08it/s]
  train:  95%|#########5| 1142/1200 [00:08<00:00, 128.86it/s]
  train:  96%|#########6| 1156/1200 [00:08<00:00, 130.76it/s]
  train:  98%|#########7| 1170/1200 [00:08<00:00, 132.42it/s]
  train:  99%|#########8| 1184/1200 [00:08<00:00, 131.78it/s]
  train: 100%|#########9| 1198/1200 [00:08<00:00, 125.32it/s]
  train: 100%|##########| 1200/1200 [00:08<00:00, 134.27it/s]
+
+Copying 150 files to val...
+
  val:   0%|          | 0/150 [00:00<?, ?it/s]
  val:   9%|8         | 13/150 [00:00<00:01, 122.03it/s]
  val:  17%|#7        | 26/150 [00:00<00:01, 123.63it/s]
  val:  28%|##8       | 42/150 [00:00<00:00, 138.84it/s]
  val:  37%|###7      | 56/150 [00:00<00:00, 133.35it/s]
  val:  47%|####6     | 70/150 [00:00<00:00, 131.76it/s]
  val:  57%|#####6    | 85/150 [00:00<00:00, 136.72it/s]
  val:  66%|######6   | 99/150 [00:00<00:00, 136.24it/s]
  val:  75%|#######5  | 113/150 [00:00<00:00, 129.22it/s]
  val:  85%|########4 | 127/150 [00:00<00:00, 131.40it/s]
  val:  94%|#########3| 141/150 [00:01<00:00, 131.03it/s]
  val: 100%|##########| 150/150 [00:01<00:00, 132.65it/s]
+
+Copying 150 files to test...
+
  test:   0%|          | 0/150 [00:00<?, ?it/s]
  test:  11%|#         | 16/150 [00:00<00:00, 154.96it/s]
  test:  21%|##1       | 32/150 [00:00<00:00, 142.70it/s]
  test:  31%|###1      | 47/150 [00:00<00:00, 137.28it/s]
  test:  41%|####      | 61/150 [00:00<00:00, 129.05it/s]
  test:  49%|####9     | 74/150 [00:00<00:00, 125.64it/s]
  test:  58%|#####8    | 87/150 [00:00<00:00, 124.07it/s]
  test:  67%|######7   | 101/150 [00:00<00:00, 126.78it/s]
  test:  77%|#######6  | 115/150 [00:00<00:00, 124.93it/s]
  test:  85%|########5 | 128/150 [00:01<00:00, 122.65it/s]
  test:  94%|#########3| 141/150 [00:01<00:00, 119.77it/s]
  test: 100%|##########| 150/150 [00:01<00:00, 126.80it/s]
+
+Created data.yaml: C:\Users\82104\Downloads\COCO_format\dataset\data.yaml
+
+============================================================
+Dataset split completed!
+============================================================
+Output directory: C:\Users\82104\Downloads\COCO_format\dataset
+
+Structure:
+  dataset/
+    ������ images/
+    ��   ������ train/  (1200 images)
+    ��   ������ val/    (150 images)
+    ��   ������ test/   (150 images)
+    ������ labels/
+    ��   ������ train/  (1200 labels)
+    ��   ������ val/    (150 labels)
+    ��   ������ test/   (150 labels)
+    ������ data.yaml