不调整大小的 DataGenerator(Sequence)

Question

我创建了一个DataGenerator(Sequence)定义batch_size,batch_x和 的类batch_y。batch_x是一批图像（来自x_set，图像的文件路径列表），由 读入imread，调整大小resize并除以 255 以获得 0 到 1 之间的值。batch_y这些图像的标签来自y_set，a包含所有标签的列表。class DataGenerator(Sequence):    def __init__(self, x_set, y_set, batch_size):        self.x, self.y = x_set, y_set        self.batch_size = batch_size    def __len__(self):        return math.ceil(len(self.x) / self.batch_size)    def __getitem__(self, idx):        batch_x = self.x[idx*self.batch_size : (idx + 1)*self.batch_size]        batch_x = np.array([resize(imread(file_name), (64, 128)) for file_name in batch_x])        batch_x = batch_x * 1./255        batch_y = self.y[idx*self.batch_size : (idx + 1)*self.batch_size]        batch_y = np.array(batch_y)        return batch_x, batch_y因为这个生成器可以工作但在 Colab 上需要很长时间，所以我之前调整了图像的大小。因此，这不再是必需的，我现在想修改DataGenerator并保留该resize功能。这是我的代码DataGenerator_withoutresize(Sequence)：class DataGenerator_withoutresize(Sequence):    def __init__(self, x_set, y_set, batch_size):        self.x, self.y = x_set, y_set        self.batch_size = batch_size    def __len__(self):        return math.ceil(len(self.x) / self.batch_size)    def __getitem__(self, idx):        batch_x = self.x[idx*self.batch_size : (idx + 1)*self.batch_size]        batch_x = np.array([(imread(file_name) for file_name in batch_x])        batch_x = batch_x * 1./255        batch_y = self.y[idx*self.batch_size : (idx + 1)*self.batch_size]        batch_y = np.array(batch_y)        return batch_x, batch_y这段代码正确吗？

Answer 1

最后，我使用了这段代码，它对我有用：

class DataGenerator(Sequence):

    def __init__(self, x_set, y_set, batch_size):

        self.x, self.y = x_set, y_set

        self.batch_size = batch_size

    def __len__(self):

        return math.ceil(len(self.x) / self.batch_size)

    def __getitem__(self, idx):

        batch_x = self.x[idx*self.batch_size : (idx + 1)*self.batch_size]

        batch_x = [imread(file_name) for file_name in batch_x]

        batch_x = np.array(batch_x)

        batch_x = batch_x * 1./255

        batch_y = self.y[idx*self.batch_size : (idx + 1)*self.batch_size]

        batch_y = np.array(batch_y)

        return batch_x, batch_y