File size: 15,697 Bytes
8eab354 |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 |
"""
Evaluation metrics for signature verification.
"""
import torch
import numpy as np
from typing import List, Tuple, Dict, Optional
from sklearn.metrics import (
accuracy_score, precision_score, recall_score, f1_score,
roc_auc_score, roc_curve, precision_recall_curve, confusion_matrix
)
import matplotlib.pyplot as plt
import seaborn as sns
class SignatureVerificationMetrics:
"""
Comprehensive metrics for signature verification evaluation.
"""
def __init__(self, threshold: float = 0.5):
"""
Initialize metrics calculator.
Args:
threshold: Similarity threshold for binary classification
"""
self.threshold = threshold
self.reset()
def reset(self):
"""Reset all stored predictions and labels."""
self.predictions = []
self.labels = []
self.similarities = []
def update(self,
similarities: np.ndarray,
labels: np.ndarray):
"""
Update metrics with new predictions.
Args:
similarities: Similarity scores
labels: Ground truth labels (1 for genuine, 0 for forged)
"""
self.similarities.extend(similarities)
self.labels.extend(labels)
# Convert similarities to binary predictions
predictions = (similarities >= self.threshold).astype(int)
self.predictions.extend(predictions)
def compute_metrics(self) -> Dict[str, float]:
"""
Compute all evaluation metrics.
Returns:
Dictionary of metrics
"""
if not self.predictions or not self.labels:
raise ValueError("No predictions or labels available. Call update() first.")
similarities = np.array(self.similarities)
labels = np.array(self.labels)
predictions = np.array(self.predictions)
# Basic classification metrics
accuracy = accuracy_score(labels, predictions)
precision = precision_score(labels, predictions, zero_division=0)
recall = recall_score(labels, predictions, zero_division=0)
f1 = f1_score(labels, predictions, zero_division=0)
# ROC AUC
try:
roc_auc = roc_auc_score(labels, similarities)
except ValueError:
roc_auc = 0.0
# Precision-Recall AUC
try:
precision_vals, recall_vals, _ = precision_recall_curve(labels, similarities)
pr_auc = np.trapz(precision_vals, recall_vals)
except ValueError:
pr_auc = 0.0
# Confusion matrix
cm = confusion_matrix(labels, predictions)
tn, fp, fn, tp = cm.ravel() if cm.size == 4 else (0, 0, 0, 0)
# Additional metrics
specificity = tn / (tn + fp) if (tn + fp) > 0 else 0.0
sensitivity = tp / (tp + fn) if (tp + fn) > 0 else 0.0
# Equal Error Rate (EER)
eer = self._compute_eer(labels, similarities)
# False Acceptance Rate (FAR) and False Rejection Rate (FRR)
far = fp / (fp + tn) if (fp + tn) > 0 else 0.0
frr = fn / (fn + tp) if (fn + tp) > 0 else 0.0
metrics = {
'accuracy': accuracy,
'precision': precision,
'recall': recall,
'f1_score': f1,
'roc_auc': roc_auc,
'pr_auc': pr_auc,
'specificity': specificity,
'sensitivity': sensitivity,
'eer': eer,
'far': far,
'frr': frr,
'threshold': self.threshold
}
return metrics
def _compute_eer(self, labels: np.ndarray, similarities: np.ndarray) -> float:
"""
Compute Equal Error Rate (EER).
Args:
labels: Ground truth labels
similarities: Similarity scores
Returns:
Equal Error Rate
"""
try:
fpr, tpr, thresholds = roc_curve(labels, similarities)
fnr = 1 - tpr
eer_threshold = thresholds[np.nanargmin(np.absolute((fnr - fpr)))]
eer = fpr[np.nanargmin(np.absolute((fnr - fpr)))]
return float(eer)
except (ValueError, IndexError):
return 0.0
def plot_roc_curve(self, save_path: Optional[str] = None):
"""
Plot ROC curve.
Args:
save_path: Path to save the plot
"""
if not self.similarities or not self.labels:
raise ValueError("No data available for plotting.")
similarities = np.array(self.similarities)
labels = np.array(self.labels)
fpr, tpr, _ = roc_curve(labels, similarities)
roc_auc = roc_auc_score(labels, similarities)
plt.figure(figsize=(8, 6))
plt.plot(fpr, tpr, color='darkorange', lw=2, label=f'ROC curve (AUC = {roc_auc:.2f})')
plt.plot([0, 1], [0, 1], color='navy', lw=2, linestyle='--')
plt.xlim([0.0, 1.0])
plt.ylim([0.0, 1.05])
plt.xlabel('False Positive Rate')
plt.ylabel('True Positive Rate')
plt.title('Receiver Operating Characteristic (ROC) Curve')
plt.legend(loc="lower right")
plt.grid(True)
if save_path:
plt.savefig(save_path, dpi=300, bbox_inches='tight')
plt.show()
def plot_precision_recall_curve(self, save_path: Optional[str] = None):
"""
Plot Precision-Recall curve.
Args:
save_path: Path to save the plot
"""
if not self.similarities or not self.labels:
raise ValueError("No data available for plotting.")
similarities = np.array(self.similarities)
labels = np.array(self.labels)
precision, recall, _ = precision_recall_curve(labels, similarities)
pr_auc = np.trapz(precision, recall)
plt.figure(figsize=(8, 6))
plt.plot(recall, precision, color='darkorange', lw=2, label=f'PR curve (AUC = {pr_auc:.2f})')
plt.xlim([0.0, 1.0])
plt.ylim([0.0, 1.05])
plt.xlabel('Recall')
plt.ylabel('Precision')
plt.title('Precision-Recall Curve')
plt.legend(loc="lower left")
plt.grid(True)
if save_path:
plt.savefig(save_path, dpi=300, bbox_inches='tight')
plt.show()
def plot_confusion_matrix(self, save_path: Optional[str] = None):
"""
Plot confusion matrix.
Args:
save_path: Path to save the plot
"""
if not self.predictions or not self.labels:
raise ValueError("No data available for plotting.")
cm = confusion_matrix(self.labels, self.predictions)
plt.figure(figsize=(8, 6))
sns.heatmap(cm, annot=True, fmt='d', cmap='Blues',
xticklabels=['Forged', 'Genuine'],
yticklabels=['Forged', 'Genuine'])
plt.title('Confusion Matrix')
plt.xlabel('Predicted')
plt.ylabel('Actual')
if save_path:
plt.savefig(save_path, dpi=300, bbox_inches='tight')
plt.show()
def plot_similarity_distribution(self, save_path: Optional[str] = None):
"""
Plot distribution of similarity scores for genuine and forged pairs.
Args:
save_path: Path to save the plot
"""
if not self.similarities or not self.labels:
raise ValueError("No data available for plotting.")
similarities = np.array(self.similarities)
labels = np.array(self.labels)
genuine_similarities = similarities[labels == 1]
forged_similarities = similarities[labels == 0]
plt.figure(figsize=(10, 6))
plt.hist(genuine_similarities, bins=50, alpha=0.7, label='Genuine', color='green')
plt.hist(forged_similarities, bins=50, alpha=0.7, label='Forged', color='red')
plt.axvline(self.threshold, color='black', linestyle='--', label=f'Threshold = {self.threshold}')
plt.xlabel('Similarity Score')
plt.ylabel('Frequency')
plt.title('Distribution of Similarity Scores')
plt.legend()
plt.grid(True, alpha=0.3)
if save_path:
plt.savefig(save_path, dpi=300, bbox_inches='tight')
plt.show()
class ThresholdOptimizer:
"""
Optimize threshold for signature verification.
"""
def __init__(self, metric: str = 'f1_score'):
"""
Initialize threshold optimizer.
Args:
metric: Metric to optimize ('f1_score', 'accuracy', 'eer')
"""
self.metric = metric
self.best_threshold = 0.5
self.best_score = 0.0
def optimize(self,
similarities: np.ndarray,
labels: np.ndarray,
threshold_range: Tuple[float, float] = (0.0, 1.0),
num_thresholds: int = 100) -> Dict[str, float]:
"""
Optimize threshold for given metric.
Args:
similarities: Similarity scores
labels: Ground truth labels
threshold_range: Range of thresholds to test
num_thresholds: Number of thresholds to test
Returns:
Dictionary with best threshold and score
"""
thresholds = np.linspace(threshold_range[0], threshold_range[1], num_thresholds)
scores = []
for threshold in thresholds:
predictions = (similarities >= threshold).astype(int)
if self.metric == 'f1_score':
score = f1_score(labels, predictions, zero_division=0)
elif self.metric == 'accuracy':
score = accuracy_score(labels, predictions)
elif self.metric == 'eer':
# Compute EER for this threshold
fpr, tpr, _ = roc_curve(labels, similarities)
fnr = 1 - tpr
try:
score = fpr[np.nanargmin(np.absolute((fnr - fpr)))]
except (ValueError, IndexError):
score = 1.0
else:
raise ValueError(f"Unsupported metric: {self.metric}")
scores.append(score)
# Find best threshold
if self.metric == 'eer':
best_idx = np.argmin(scores)
else:
best_idx = np.argmax(scores)
self.best_threshold = thresholds[best_idx]
self.best_score = scores[best_idx]
return {
'best_threshold': self.best_threshold,
'best_score': self.best_score,
'thresholds': thresholds,
'scores': scores
}
def plot_threshold_analysis(self,
similarities: np.ndarray,
labels: np.ndarray,
save_path: Optional[str] = None):
"""
Plot threshold analysis.
Args:
similarities: Similarity scores
labels: Ground truth labels
save_path: Path to save the plot
"""
result = self.optimize(similarities, labels)
plt.figure(figsize=(10, 6))
plt.plot(result['thresholds'], result['scores'], 'b-', linewidth=2)
plt.axvline(self.best_threshold, color='red', linestyle='--',
label=f'Best threshold = {self.best_threshold:.3f}')
plt.xlabel('Threshold')
plt.ylabel(f'{self.metric.upper()}')
plt.title(f'Threshold Optimization - {self.metric.upper()}')
plt.legend()
plt.grid(True, alpha=0.3)
if save_path:
plt.savefig(save_path, dpi=300, bbox_inches='tight')
plt.show()
class CrossValidationEvaluator:
"""
Cross-validation evaluator for signature verification.
"""
def __init__(self,
model: torch.nn.Module,
k_folds: int = 5,
threshold: float = 0.5):
"""
Initialize cross-validation evaluator.
Args:
model: Model to evaluate
k_folds: Number of folds for cross-validation
threshold: Similarity threshold
"""
self.model = model
self.k_folds = k_folds
self.threshold = threshold
self.results = []
def evaluate(self,
data_pairs: List[Tuple[str, str, int]],
preprocessor,
batch_size: int = 32) -> Dict[str, float]:
"""
Perform k-fold cross-validation.
Args:
data_pairs: List of (signature1_path, signature2_path, label) tuples
preprocessor: Image preprocessor
batch_size: Batch size for evaluation
Returns:
Average metrics across all folds
"""
from sklearn.model_selection import KFold
kf = KFold(n_splits=self.k_folds, shuffle=True, random_state=42)
data_pairs = np.array(data_pairs)
fold_metrics = []
for fold, (train_idx, val_idx) in enumerate(kf.split(data_pairs)):
print(f"Evaluating fold {fold + 1}/{self.k_folds}")
val_pairs = data_pairs[val_idx]
# Evaluate on validation set
fold_metrics.append(self._evaluate_fold(val_pairs, preprocessor, batch_size))
# Compute average metrics
avg_metrics = {}
for metric in fold_metrics[0].keys():
avg_metrics[metric] = np.mean([fold[metric] for fold in fold_metrics])
avg_metrics[f'{metric}_std'] = np.std([fold[metric] for fold in fold_metrics])
self.results = fold_metrics
return avg_metrics
def _evaluate_fold(self,
val_pairs: np.ndarray,
preprocessor,
batch_size: int) -> Dict[str, float]:
"""
Evaluate a single fold.
Args:
val_pairs: Validation pairs
preprocessor: Image preprocessor
batch_size: Batch size
Returns:
Metrics for this fold
"""
self.model.eval()
similarities = []
labels = []
with torch.no_grad():
for i in range(0, len(val_pairs), batch_size):
batch_pairs = val_pairs[i:i+batch_size]
for sig1_path, sig2_path, label in batch_pairs:
# Load and preprocess images
sig1 = preprocessor.preprocess_image(sig1_path)
sig2 = preprocessor.preprocess_image(sig2_path)
# Add batch dimension
sig1 = sig1.unsqueeze(0)
sig2 = sig2.unsqueeze(0)
# Compute similarity
similarity = self.model(sig1, sig2)
similarities.append(similarity.item())
labels.append(label)
# Compute metrics
similarities = np.array(similarities)
labels = np.array(labels)
metrics_calculator = SignatureVerificationMetrics(threshold=self.threshold)
metrics_calculator.update(similarities, labels)
return metrics_calculator.compute_metrics()
|