File size: 10,530 Bytes
1f2e262 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 | #!/usr/bin/env python3
"""
Data Audit Script - Stage 3
Checks for duplicates, corruption, and dataset integrity.
"""
import os
import sys
from pathlib import Path
from collections import defaultdict
import hashlib
from PIL import Image
import imagehash
import pandas as pd
from tqdm import tqdm
# Add project root to path
project_root = Path(__file__).parent.parent.parent
sys.path.insert(0, str(project_root))
def compute_file_hash(filepath: str) -> str:
"""Compute MD5 hash of file."""
hash_md5 = hashlib.md5()
with open(filepath, "rb") as f:
for chunk in iter(lambda: f.read(4096), b""):
hash_md5.update(chunk)
return hash_md5.hexdigest()
def compute_perceptual_hash(filepath: str) -> str:
"""Compute perceptual hash using difference hash."""
try:
img = Image.open(filepath)
return str(imagehash.dhash(img, hash_size=8))
except Exception as e:
return None
def audit_dataset(base_dir: str = "Dataset"):
"""
Audit the brain tumor dataset.
Checks:
1. File counts per class and split
2. Image properties (size, channels, format)
3. Exact duplicates (file hash)
4. Near-duplicates (perceptual hash)
5. Corrupt files
6. Cross-split leakage
"""
print("="*70)
print("Brain Tumor MRI Dataset Audit")
print("="*70)
base_path = Path(base_dir)
if not base_path.exists():
print(f"✗ Dataset directory not found: {base_dir}")
print(" Expected structure: Dataset/Training/ and Dataset/Testing/")
return False
results = {
"splits": {},
"duplicates": [],
"near_duplicates": [],
"corrupt": [],
"cross_split_leakage": []
}
# Expected classes
expected_classes = ["glioma", "meningioma", "notumor", "pituitary"]
# Process each split
for split in ["Training", "Testing"]:
split_dir = base_path / split
if not split_dir.exists():
print(f"\n✗ {split} directory not found")
continue
print(f"\n{'='*70}")
print(f"Auditing {split} split...")
print(f"{'='*70}")
split_data = {
"classes": {},
"file_hashes": {},
"perceptual_hashes": defaultdict(list),
"total_files": 0
}
# Process each class
for class_name in expected_classes:
class_dir = split_dir / class_name
if not class_dir.exists():
print(f" ✗ Class directory not found: {class_name}")
continue
image_files = list(class_dir.glob("*.jpg")) + list(class_dir.glob("*.jpeg")) + list(class_dir.glob("*.png"))
class_data = {
"count": len(image_files),
"sizes": set(),
"channels": set(),
"formats": set()
}
# Analyze each image
for img_path in tqdm(image_files, desc=f" {class_name}", leave=False):
# File hash
file_hash = compute_file_hash(str(img_path))
# Check for exact duplicates within split
if file_hash in split_data["file_hashes"]:
results["duplicates"].append({
"split": split,
"file1": split_data["file_hashes"][file_hash],
"file2": str(img_path)
})
else:
split_data["file_hashes"][file_hash] = str(img_path)
# Perceptual hash
phash = compute_perceptual_hash(str(img_path))
if phash:
split_data["perceptual_hashes"][phash].append(str(img_path))
# Image properties
try:
with Image.open(img_path) as img:
class_data["sizes"].add(img.size)
class_data["channels"].add(img.mode)
class_data["formats"].add(img.format)
except Exception as e:
results["corrupt"].append({
"file": str(img_path),
"error": str(e)
})
split_data["classes"][class_name] = class_data
split_data["total_files"] += class_data["count"]
# Print class summary
print(f" {class_name}: {class_data['count']} images")
if class_data["sizes"]:
size_summary = list(class_data["sizes"])[:5]
print(f" Sizes (sample): {size_summary}")
if class_data["channels"]:
print(f" Channels: {class_data['channels']}")
results["splits"][split] = split_data
print(f"\n Total {split} images: {split_data['total_files']}")
# Check for near-duplicates within each split
print(f"\n{'='*70}")
print("Checking for near-duplicates...")
print(f"{'='*70}")
for split, split_data in results["splits"].items():
for phash, file_list in split_data["perceptual_hashes"].items():
if len(file_list) > 1:
results["near_duplicates"].append({
"split": split,
"files": file_list,
"hash": phash
})
# Check for cross-split leakage
print("\nChecking for cross-split leakage...")
if "Training" in results["splits"] and "Testing" in results["splits"]:
train_hashes = set(results["splits"]["Training"]["perceptual_hashes"].keys())
test_hashes = set(results["splits"]["Testing"]["perceptual_hashes"].keys())
common_hashes = train_hashes & test_hashes
for phash in common_hashes:
train_files = results["splits"]["Training"]["perceptual_hashes"][phash]
test_files = results["splits"]["Testing"]["perceptual_hashes"][phash]
results["cross_split_leakage"].append({
"hash": phash,
"train_files": train_files,
"test_files": test_files
})
# Print summary
print(f"\n{'='*70}")
print("Audit Summary")
print(f"{'='*70}")
# Dataset size
expected_train = 5600
expected_test = 1600
actual_train = results["splits"].get("Training", {}).get("total_files", 0)
actual_test = results["splits"].get("Testing", {}).get("total_files", 0)
print(f"\nDataset size:")
print(f" Training: {actual_train} (expected {expected_train})")
print(f" Testing: {actual_test} (expected {expected_test})")
if actual_train == expected_train and actual_test == expected_test:
print(" ✓ File counts match expected")
else:
print(" ⚠ File counts do not match expected")
# Issues found
issues_found = False
if results["duplicates"]:
print(f"\n✗ Exact duplicates found: {len(results['duplicates'])}")
for dup in results["duplicates"][:5]:
print(f" {dup['file1']}")
print(f" {dup['file2']}")
issues_found = True
else:
print("\n✓ No exact duplicates found")
if results["near_duplicates"]:
print(f"\n⚠ Near-duplicates found: {len(results['near_duplicates'])} groups")
for i, dup_group in enumerate(results["near_duplicates"][:3]):
print(f" Group {i+1} ({len(dup_group['files'])} files):")
for f in dup_group['files'][:2]:
print(f" {f}")
else:
print("✓ No near-duplicates found")
if results["cross_split_leakage"]:
print(f"\n✗ Cross-split leakage found: {len(results['cross_split_leakage'])} cases")
print(" ⚠ Training and testing sets contain similar images!")
for i, leak in enumerate(results["cross_split_leakage"][:3]):
print(f" Case {i+1}:")
print(f" Train: {leak['train_files'][0]}")
print(f" Test: {leak['test_files'][0]}")
issues_found = True
else:
print("✓ No cross-split leakage detected")
if results["corrupt"]:
print(f"\n✗ Corrupt files found: {len(results['corrupt'])}")
for corrupt in results["corrupt"][:5]:
print(f" {corrupt['file']}: {corrupt['error']}")
issues_found = True
else:
print("✓ No corrupt files found")
# Save detailed report
report_dir = Path("reports")
report_dir.mkdir(exist_ok=True)
report_path = report_dir / "audit_report.txt"
with open(report_path, "w") as f:
f.write("Brain Tumor MRI Dataset Audit Report\n")
f.write("="*70 + "\n\n")
f.write(f"Training images: {actual_train}\n")
f.write(f"Testing images: {actual_test}\n\n")
f.write(f"Exact duplicates: {len(results['duplicates'])}\n")
f.write(f"Near-duplicates: {len(results['near_duplicates'])}\n")
f.write(f"Cross-split leakage: {len(results['cross_split_leakage'])}\n")
f.write(f"Corrupt files: {len(results['corrupt'])}\n\n")
if results["cross_split_leakage"]:
f.write("\nCross-split leakage details:\n")
for leak in results["cross_split_leakage"]:
f.write(f"\nHash: {leak['hash']}\n")
f.write(f" Training: {leak['train_files']}\n")
f.write(f" Testing: {leak['test_files']}\n")
print(f"\nDetailed report saved to: {report_path}")
print(f"\n{'='*70}")
if not issues_found:
print("✓ Dataset audit passed - ready for preprocessing")
return True
else:
print("⚠ Issues found - review before proceeding")
print("\nRecommendations:")
if results["duplicates"] or results["cross_split_leakage"]:
print(" - Remove duplicates from Training set (keep Testing untouched)")
if results["corrupt"]:
print(" - Remove or fix corrupt files")
return False
def main():
"""Main audit routine."""
try:
success = audit_dataset()
sys.exit(0 if success else 1)
except Exception as e:
print(f"\n✗ Audit failed with error: {e}")
import traceback
traceback.print_exc()
sys.exit(1)
if __name__ == "__main__":
main()
|