TypeError：预期的二进制或Unicode字符串

Question

我想使用tf.data输入我的图像数据。我已经将所有图像折叠成一个np.array，然后我习惯了np.array创建一个tf.data.Dataset对象。但是，我遇到了TypeError。我的代码如下所示。

import os
from scipy.misc import imread
import numpy as np
import glob
import tensorflow as tf

os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2'
image = []
img_dir = 'data/ILSVRC2012_test/*'

images = np.array([np.array(imread(data)) for data in glob.glob(img_dir)])

image_data = tf.data.Dataset.from_tensor_slices(images)

下面的块是错误信息。

TypeError: Expected binary or unicode string, got array([[[184, 210, 225],
        [184, 210, 225],
        [184, 210, 225],
        ...,
        [160, 185, 205],
        [159, 184, 204],
        [159, 184, 204]],

       [[183, 209, 224],
        [184, 210, 225],
        [184, 210, 225],
        ...,
        [159, 186, 205],
        [159, 186, 205],
        [159, 186, 205]],

       [[184, 210, 225],
        [184, 210, 225],
        [185, 211, 226],
        ...,
        [160, 187, 206],
        [160, 187, 206],
        [160, 187, 206]],

       ...,

       [[ 65,  65,  15],
        [ 71,  71,  17],
        [ 75,  76,  19],
        ...,
        [ 83,  83,  19],
        [ 82,  87,  21],
        [ 85,  85,  21]],

       [[ 70,  70,  18],
        [ 74,  75,  18],
        [ 74,  78,  19],
        ...,
        [ 77,  81,  20],
        [ 78,  87,  24],
        [ 77,  81,  20]],

       [[ 71,  71,  17],
        [ 73,  74,  17],
        [ 77,  78,  20],
        ...,
        [ 85,  86,  20],
        [ 85,  85,  21],
        [ 75,  74,  20]]], dtype=uint8)

任何帮助将不胜感激！