accuracy = num_correct * 100.0 / len(test_images_filenames)
# Show results and timing
print('\nACCURACY: {:.2f}'.format(accuracy))
print('\nTOTAL TIME: {:.2f} s'.format(time.time() - start))
classes = lin_svm.classes_
# Create confusion matrix
conf = confusion_matrix(test_labels, pred_class, labels=classes)
# Create ROC curve and AUC score
test_labels_bin = label_binarize(test_labels, classes=classes)
fpr = []
tpr = []
roc_auc = []
for i in range(len(classes)):
c_fpr, c_tpr, _ = roc_curve(test_labels_bin[:, i], np.array(pred_prob)[:, i])
c_roc_auc = auc(c_fpr, c_tpr)
fpr.append(c_fpr)
tpr.append(c_tpr)
roc_auc.append(c_roc_auc)
# Plot
plotting.plot_confusion_matrix(conf, classes=classes, normalize=True)
plotting.plot_roc_curve(fpr, tpr, roc_auc, classes=classes,
title='ROC curve for linear SVM with codebook of {} words'.format(settings.codebook_size)
)
print('Done.')
test_results = io.load_object('svm_denseSIFT_test_results', ignore=True)
classes = io.load_object('svm_denseSIFT_classes', ignore=True)
pred_results = [x[0] for x in test_results]
pred_class = [x[1] for x in test_results]
pred_prob = [x[2] for x in test_results]
test_images_filenames, test_labels = io.load_test_set()
# Create confusion matrix
conf = confusion_matrix(test_labels, pred_class, labels=classes)
# Create ROC curve and AUC score
test_labels_bin = label_binarize(test_labels, classes=classes)
fpr = []
tpr = []
roc_auc = []
for i in range(len(classes)):
c_fpr, c_tpr, _ = roc_curve(test_labels_bin[:, i], np.array(pred_prob)[:, i])
c_roc_auc = auc(c_fpr, c_tpr)
fpr.append(c_fpr)
tpr.append(c_tpr)
roc_auc.append(c_roc_auc)
# Plot
plotting.plot_confusion_matrix(conf, classes=classes, normalize=True)
plotting.plot_roc_curve(fpr, tpr, roc_auc, classes=classes,
title='ROC curve for rbf SVM with codebook of {} words'.format(K)
)
print('Done.')
