def test_precision_recall_curve():
bin_class = BinaryClassificationMetrics(cls='a', FN=FN, TP=TP, FP=FP)
metrics = BinaryDetectionMetrics(bin_class, 10, predictions)
rounded_pr_curve = [(round(rp[0], 4), round(rp[1], 4))
for rp in metrics.precision_recall_curve]
expected_curve = [
(0.0667, 1.0000),
(0.0667, 0.5000),
(0.1333, 0.6667),
(0.1333, 0.5000),
(0.1333, 0.4000),
(0.1333, 0.3333),
(0.1333, 0.2857),
(0.1333, 0.2500),
(0.1333, 0.2222),
(0.2000, 0.3000),
(0.2000, 0.2727),
(0.2667, 0.3333),
(0.3333, 0.3846),
(0.4000, 0.4286),
(0.4000, 0.4000),
(0.4000, 0.3750),
(0.4000, 0.3529),
(0.4000, 0.3333),
(0.4000, 0.3158),
(0.4000, 0.3000),
(0.4000, 0.2857),
(0.4000, 0.2727),
(0.4667, 0.3043),
(0.4667, 0.2917),
]
assert rounded_pr_curve == expected_curve
def test_pred_count():
bin_class = BinaryClassificationMetrics(cls='a', FN=FN, TP=TP, FP=FP)
metrics = BinaryDetectionMetrics(bin_class, 0, predictions)
assert metrics.predicted_count == 24
def test_iou():
bin_class = BinaryClassificationMetrics(cls='a', FN=FN, TP=TP, FP=FP)
metrics = BinaryDetectionMetrics(bin_class, 10, predictions)
assert metrics.iou == 10
def test_average_precision():
bin_class = BinaryClassificationMetrics(cls='a', FN=FN, TP=TP, FP=FP)
metrics = BinaryDetectionMetrics(bin_class, 10, predictions)
assert round(metrics.average_precision(11), 4) == .2684
assert round(metrics.average_precision(), 4) == .2457