Datasets:
File size: 6,155 Bytes
abcc75c | 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 | import numpy as np
import json
import os
import sys
def calculate_precision_at_1(predictions, ground_truth):
"""
Calculate precision@1 for the given predictions and ground truth.
Args:
predictions (np.array): Array of predicted gallery indices for each query
ground_truth (np.array): Array of correct gallery indices for each query
Returns:
float: Precision@1 score (percentage of correct predictions)
"""
if len(predictions) != len(ground_truth):
raise ValueError(f"Predictions length ({len(predictions)}) doesn't match ground truth length ({len(ground_truth)})")
# Count correct predictions
correct_predictions = np.sum(predictions == ground_truth)
total_predictions = len(predictions)
# Calculate precision@1 as percentage
precision_at_1 = (correct_predictions / total_predictions)
return precision_at_1
def load_submission_file(filepath):
"""
Load submission file and handle potential errors.
Args:
filepath (str): Path to the submission file
Returns:
np.array or None: Loaded array or None if file doesn't exist or is invalid
"""
try:
if not os.path.exists(filepath):
print(f"Warning: Submission file {filepath} not found")
return None
submission = np.load(filepath)
print(f"Loaded {filepath}: shape {submission.shape}, dtype {submission.dtype}")
return submission
except Exception as e:
print(f"Error loading {filepath}: {str(e)}")
return None
def load_ground_truth_file(filepath):
"""
Load ground truth file.
Args:
filepath (str): Path to the ground truth file
Returns:
np.array: Loaded ground truth array
"""
try:
if not os.path.exists(filepath):
raise FileNotFoundError(f"Ground truth file {filepath} not found")
ground_truth = np.load(filepath)
print(f"Loaded ground truth {filepath}: shape {ground_truth.shape}, dtype {ground_truth.dtype}")
return ground_truth
except Exception as e:
print(f"Error loading ground truth {filepath}: {str(e)}")
raise
def evaluate_test_set(submission_file, ground_truth_file, test_name):
"""
Evaluate a single test set.
Args:
submission_file (str): Path to submission file
ground_truth_file (str): Path to ground truth file
test_name (str): Name of the test set for logging
Returns:
float or None: Precision@1 score or None if evaluation failed
"""
print(f"\n=== Evaluating {test_name} ===")
# Load ground truth
try:
ground_truth = load_ground_truth_file(ground_truth_file)
except Exception as e:
print(f"Failed to load ground truth for {test_name}: {str(e)}")
return None
# Load submission
submission = load_submission_file(submission_file)
if submission is None:
print(f"Failed to load submission for {test_name}")
return None
# Validate submission format
if submission.shape != ground_truth.shape:
print(f"Shape mismatch for {test_name}: submission {submission.shape} vs ground truth {ground_truth.shape}")
return None
# Calculate precision@1
try:
score = calculate_precision_at_1(submission, ground_truth)
print(f"{test_name} - Precision@1: {score:.2f}")
# Log some statistics
correct_count = np.sum(submission == ground_truth)
total_count = len(submission)
print(f"{test_name} - Correct predictions: {correct_count}/{total_count}")
return score
except Exception as e:
print(f"Error calculating precision@1 for {test_name}: {str(e)}")
return None
def main():
"""
Main evaluation function.
"""
print("Starting evaluation...")
if os.environ.get('METRIC_PATH'):
METRIC_PATH = os.environ.get("METRIC_PATH") + "/"
else:
METRIC_PATH = "" # Fallback for local testing
# File paths
submission_a_file = "submission_a.npy"
submission_b_file = "submission_b.npy"
ground_truth_a_file = METRIC_PATH + "answer_a.npy"
ground_truth_b_file = METRIC_PATH + "answer_b.npy"
output_file = "score.json"
# Evaluate test set A
score_a = evaluate_test_set(submission_a_file, ground_truth_a_file, "Test Set A")
# Evaluate test set B
score_b = evaluate_test_set(submission_b_file, ground_truth_b_file, "Test Set B")
# Determine overall status
status = True
msg = "Success!"
# Handle missing or failed evaluations
if score_a is None:
score_a = 0.0
status = False
msg = "Failed to evaluate Test Set A"
if score_b is None:
score_b = 0.0
if status: # Only update if not already failed
status = False
msg = "Failed to evaluate Test Set B"
else:
msg = "Failed to evaluate both test sets"
if score_a > 1:
score_a = 0.0
if score_b > 1:
score_b = 0.0
def sanitize_score(value):
"""处理单个分数值，将NaN和inf替换为0"""
if not np.isfinite(value):
return 0.0
return value
# Create result dictionary
result = {
"status": status,
"score": {
"public_a": sanitize_score(score_a),
"private_b": sanitize_score(score_b),
},
"msg": msg,
}
# Save results to JSON
try:
with open(output_file, 'w') as f:
json.dump(result, f, indent=4)
print(f"\nResults saved to {output_file}")
except Exception as e:
print(f"Error saving results to {output_file}: {str(e)}")
sys.exit(1)
# Print final summary
print("\n=== EVALUATION SUMMARY ===")
print(f"Status: {status}")
print(f"Test Set A (public) Score: {score_a:.2f}")
print(f"Test Set B (private) Score: {score_b:.2f}")
print(f"Message: {msg}")
return result
if __name__ == "__main__":
main()
|