将train/test/validation中的运动图像拆分到正确目录时出现问题

Question

我正在尝试运行一个基于Keras/Colab教程的图像分类器。以前，我手动将图像分割到train/test/validation文件夹中，效果很好。现在，当我把所有的图像放到基本目录中并拆分它们时，我遇到了问题

我想知道在使用numpy将图像分割成train、test、validation集之后，是否需要将图像直接复制到适当的目录

base_dir = 'drive/Colab/EV3/Base/'
like = os.path.join(base_dir, 'Like')
dislike = os.path.join(base_dir, 'Dislike')

trn_dir = 'drive/Colab/EV3/Train/'
val_dir = 'drive/Colab/EV3/Valid/'
tst_dir = 'drive/Colab/EV3/Test/'

trn_like = os.path.join(trn_dir, 'Like')
val_like = os.path.join(val_dir, 'Like')
tst_like = os.path.join(tst_dir, 'Like')

trn_disl = os.path.join(trn_dir, 'Dislike')
val_disl = os.path.join(val_dir, 'Dislike')
tst_disl = os.path.join(tst_dir, 'Dislike')

like_fnames.sort ()   
np.random.shuffle(like_fnames)

disl_fnames = os.listdir(dislike)
disl_fnames.sort () 
np.random.shuffle(disl_fnames)

trn_like, val_like, tst_like = np.split(like_fnames, [int(.7*len(like_fnames)), int(.85*len(like_fnames))])

trn_disl, val_disl, tst_disl = np.split(disl_fnames, [int(.7*len(disl_fnames)), int(.85*len(disl_fnames))])

image_size = 640 
batch_size = 112

train_datagen = keras.preprocessing.image.ImageDataGenerator(
                rescale=1./255)

validation_datagen = keras.preprocessing.image.ImageDataGenerator(
                rescale=1./255)

train_generator = train_datagen.flow_from_directory(
                trn_dir,  
                target_size=(image_size, image_size),
                batch_size=batch_size,
                class_mode='binary')

validation_generator = validation_datagen.flow_from_directory(
                val_dir,
                target_size=(image_size, image_size),
                batch_size=batch_size,
                class_mode='binary')

主要问题发生在最后一步，我应该得到一个输出：

Found 560 images belonging to 2 classes.
Found 120 images belonging to 2 classes.

但我得到的却是：

Found 0 images belonging to 2 classes.
Found 0 images belonging to 2 classes.

如果我犯了重大的，明显的错误，我道歉。我对python和代码非常陌生，我正在尽我所能地遵循答案和教程