# for img0 in image.cpu().numpy():
# cv2.imshow('image', img0)
# cv2.waitKey(100)
# print(label.cpu().numpy())
total += float(labels2.size(0))
correct += float(predicted.eq(labels2).sum().item())
if batch_idx % 100 == 0:
acc = 100. * float(correct) / float(total)
print(batch_idx, len(test_loader), ' Acc: %.5f' % acc)
class_names = ['0 ', '1 ', '2 ', '3 ', '4 ', '5 ', '6 ', '7 ', '8 ', '9 ']
# ['0_zoom_out', '1_zoom_out', '2_zoom_out', '3_zoom_out', '4_zoom_out',
# '5_zoom_out', '6_zoom_out', '7_zoom_out', '8_zoom_out', '9_zoom_out',
# '0_zoom_in', '1_zoom_in', '2_zoom_in', '3_zoom_in', '4_zoom_in',
# '5_zoom_in', '6_zoom_in', '7_zoom_in', '8_zoom_in', '9_zoom_in']
plot_confusion_matrix(cm, class_names)
print('Iters:', epoch, '\n\n\n')
print('Test Accuracy of the model on the 10000 test images: %.3f' %
(100 * correct / total))
acc = 100. * float(correct) / float(total)
acc_record.append(acc)
if epoch % 5 == 0:
print(acc)
print('Saving..')
state = {
'net': snn.state_dict(),
'acc': acc,
'epoch': epoch,
'acc_record': acc_record,
}
if not os.path.isdir('checkpoint'):
conf_mat_train = confusion_matrix(y_train,
pred_train,
labels=[0, 1, 2, 3, 4, 5])
classes = [
'walking', 'walking up', 'walking down', 'sitting', 'standing', 'laying'
]
classes = np.array(classes)
### Printing and plotting operation of results ###
# Test accuracy
print('Test Accuracy: ', accuracy_score(y_test, pred_test))
plt.figure(1)
plot_confusion_matrix(conf_mat_test,
classes,
normalize=True,
title='Confusion matrix')
fname_test = os.getcwd() + '/Confusion Matrices/LDA_test_all.png'
plt.savefig(fname_test, dpi=256, edgecolor='b', format='png', frameon=True)
# calculating train accuracy score
print('Train Accuracy: ', accuracy_score(y_train, pred_train))
plt.figure(2)
plot_confusion_matrix(conf_mat_train,
classes,
normalize=True,
title='Confusion matrix')
fname = os.getcwd() + '/Confusion Matrices/LDA_train_all.png'
plt.savefig(fname, dpi=256, edgecolor='b', format='png', frameon=True)
plt.show()
