def get_data(): X = [] y = [] basePath = join(dir_path, "..", "common", "data", "GTSRB_subset_2") imageFileNamesClass1 = absolute_file_paths([join(basePath, "class1")]) for imgPath in imageFileNamesClass1: X.append(io.imread(imgPath)) y.append(0) imageFileNamesClass2 = absolute_file_paths([join(basePath, "class2")]) for imgPath in imageFileNamesClass2: X.append(io.imread(imgPath)) y.append(1) X = (X - np.min(X)) / np.max(X) y = to_categorical(y, 2) return X, y
def save_test_data(test_path, test_targets_path, test_coords_path, test_shape_path, orig_test_dir, orig_test_targets_dir): """Loads, formats, and re-saves test data from original directories.""" print('in save_test_data') # Gets original data files. test_files = sorted(absolute_file_paths(orig_test_dir)) test_targets_files = sorted(absolute_file_paths(orig_test_targets_dir)) # Loads and preprocesses data. test, test_coords, test_shape = load_data(test_files) test_targets, _, _ = load_data(test_targets_files, norm=False) # Re-saves data in specified directories. np.save(test_path, test) np.save(test_targets_path, test_targets) with open(test_coords_path, "wb") as a, open(test_shape_path, "wb") as b: pickle.dump(test_coords, a) pickle.dump(test_shape, b) return test, test_targets, test_coords, test_shape
def save_train_data(train_path, valid_path, train_targets_path, valid_targets_path, orig_train_dir, orig_valid_dir, orig_train_targets_dir, orig_valid_targets_dir): """Loads, formats, and re-saves train data from original directories.""" print('in save_train_data') # Gets original data files. train_files = sorted(absolute_file_paths(orig_train_dir)) valid_files = sorted(absolute_file_paths(orig_valid_dir)) train_targets_files = sorted(absolute_file_paths(orig_train_targets_dir)) valid_targets_files = sorted(absolute_file_paths(orig_valid_targets_dir)) # Loads and preprocesses data. train, _, _ = load_data(train_files) valid, _, _ = load_data(valid_files) train_targets, _, _ = load_data(train_targets_files, norm=False) valid_targets, _, _ = load_data(valid_targets_files, norm=False) # Re-saves data in specified directories. np.save(train_path, train) np.save(valid_path, valid) np.save(train_targets_path, train_targets) np.save(valid_targets_path, valid_targets) return train, valid, train_targets, valid_targets