def test_model(model_save_dir_path):
print('Started')
(model_name, channel_numbers, img_size, epochs, batch_size, starts_neuron,
start_case_index_train, end_case_index_train, start_case_index_test,
end_case_index_test) = load_variables()
print('Variables loaded')
model = get_model_builder(model_name)(img_size, img_size, channel_numbers,
starts_neuron)
print("Model built")
model.summary()
print("Model summary")
test_images, test_labels = get_images_and_masks(img_size, img_size,
start_case_index_test,
end_case_index_test, True)
print("Test data loaded")
loss, acc = model.evaluate(test_images, test_labels, verbose=1)
print("Restored model, accuracy: {:5.2f}%".format(100 * acc))
model.load_weights(model_save_dir_path)
print("Model weights loaded")
print("Prediction started")
preds_test = model.predict(test_images, verbose=1)
preds_test_t = (preds_test > 0.6).astype(np.uint8)
f1_score_result = f1_score(test_labels.flatten().flatten(),
preds_test_t.flatten().flatten())
print('F1 score: %f' % f1_score_result)
from sklearn.metrics import f1_score, precision_score, recall_score
from models_builders import get_model_builder
from utils.read_data import get_images_and_masks
from utils.utils import load_variables
print("[LOG] Starting application")
print('[LOG] Loading variables')
(model_name, channel_numbers, img_size, epochs, batch_size, starts_neuron,
start_case_index_train, end_case_index_train, start_case_index_test,
end_case_index_test, trained_model_weights_path) = load_variables()
print('[LOG] Did load variables')
print('[LOG] Building model')
model = get_model_builder(model_name)(img_size, img_size, channel_numbers,
starts_neuron)
print('[LOG] Load trained model weights')
model.load_weights(trained_model_weights_path)
print('[LOG] Loading test data for cases from: ', start_case_index_test,
" to: ", end_case_index_test)
test_images, test_labels = get_images_and_masks(img_size, img_size,
start_case_index_test,
end_case_index_test, True)
print("[LOG] Did load test data")
print("[LOG] Start model evaluation with test data")
loss, acc = model.evaluate(test_images, test_labels, verbose=1)
print("[LOG] Did evaluate model")
print("[LOG] Current model accuracy: {:5.2f}%".format(100 * acc))