def report_accuracy(X, y, coarse=True, prefix_string='benchmark'):
if coarse:
Y = np_utils.to_categorical(y, nb_classes_coarse)
else:
Y = np_utils.to_categorical(y, nb_classes_fine)
# Test the model
Y_predict = model.predict(X, batch_size=batch_size, verbose=1)
# Convert floating point vector to a clean binary vector with only two 1's
Y_predict_clean = clean_vec(Y_predict)
acc = accuracy(Y_predict_clean, Y)
print("%s accuracy: %f" % (prefix_string, acc))
Y_test_fine = np_utils.to_categorical(y_test_fine, nb_classes_fine)
Y_test_coarse = np_utils.to_categorical(y_test_coarse, nb_classes_coarse)
if 'coarse' in model_name:
Y_train = Y_train_coarse
Y_test = Y_test_coarse
elif 'fine' in model_name:
Y_train = Y_train_fine
Y_test = Y_test_fine
# Test the model
Y_predict_test = model.predict(X_test, batch_size=batch_size, verbose=1)
Y_predict_train = model.predict(X_train, batch_size=batch_size, verbose=1)
#print(Y_predict_train)
# Convert floating point vector to a clean binary vector with only two 1's
Y_predict_test_clean = clean_vec(Y_predict_test)
Y_predict_train_clean = clean_vec(Y_predict_train)
if 'coarse' in model_name:
label_type = 'coarse'
elif 'fine' in model_name:
label_type = 'fine'
test_accuracy = accuracy(Y_predict_test_clean, Y_test)
print(label_type + " test accuracy: %f" % test_accuracy)
train_accuracy = accuracy(Y_predict_train_clean, Y_train)
print(label_type + " train accuracy: %f" % train_accuracy)
# For generalization test, show the accuracy on the things it was not trained on
if '_gen' in model_name:
