将FER数据集使用Pickle处理-Python代码

之前写过一篇文章，是将FER2013数据集还原成灰度图像的Python代码，便于我们查看。

现在我想把它转换成pickle进行存储，这样便于读取和传输。

直接上代码就好了：

    def process_to_pickle(self):
        """
        将csv文件加工成pickle文件，方便存储和读取
        :return: None
        """
        # 创建文件夹
        if not os.path.exists(self.pickle_data_path): os.makedirs(self.pickle_data_path)
        # 读取csv数据
        with open(self.data_source) as f:
            train_x_data = []
            train_y_data = []
            test_x_data = []
            test_y_data = []
            validate_x_data = []
            validate_y_data = []
            # 打开文件夹
            data_csv = csv.reader(f)
            for (label, pixcels, usage) in data_csv:
                if usage == 'Training':
                    temp_list = []
                    for pixel in pixcels.split():
                        temp_list.append(pixel)
                    I = np.asarray(temp_list)
                    train_x_data.append(I.tolist())
                    train_y_data.append(label)
                elif usage == 'PrivateTest':
                    temp_list = []
                    for pixel in pixcels.split():
                        temp_list.append(pixel)
                    I = np.asarray(temp_list)
                    test_x_data.append(I.tolist())
                    test_y_data.append(label)
                elif usage == 'PublicTest':
                    temp_list = []
                    for pixel in pixcels.split():
                        temp_list.append(pixel)
                    I = np.asarray(temp_list)
                    validate_x_data.append(I.tolist())
                    validate_y_data.append(label)

            # 定义三个pickle文件
            train_pickle_file = open(os.path.join(self.pickle_data_path, 'train.txt'), 'wb')
            test_pickle_file = open(os.path.join(self.pickle_data_path, 'test.txt'), 'wb')
            validate_pickle_file = open(os.path.join(self.pickle_data_path, 'validate.txt'), 'wb')
            # 写入Pickle
            pickle.dump(train_x_data, train_pickle_file)
            pickle.dump(train_y_data, train_pickle_file)
            pickle.dump(test_x_data, test_pickle_file)
            pickle.dump(test_y_data, test_pickle_file)
            pickle.dump(validate_x_data, validate_pickle_file)
            pickle.dump(validate_y_data, validate_pickle_file)
            # 关闭文件
            train_pickle_file.close()
            test_pickle_file.close()
            validate_pickle_file.close()
            logging.info(msg='Save data finish!')

这段代码可以直接放到我的之前的文章：点这里 。里面写了一个类，直接放到里面就能够使用。

如果代码里有错误，欢迎大家指正！！