python - “ascii”编解码器无法解码位置 6 : ordinal not in range(128) 中的字节 0x8b

Question

我正在尝试使用tensorflow编写CNN，但我不断出现此错误:

UnicodeDecodeError                        Traceback (most recent call last)
<ipython-input-20-a02172d91c0c> in <module>()
     39 # Load all the data batches.
     40 for i in range(5):
---> 41     data_batch = unpickle( 'data_batch_' + str(i + 1))
     42 
     43     train_data = np.append(train_data, data_batch[b'data'])

<ipython-input-20-a02172d91c0c> in unpickle(file)
     27     import pickle
     28     with open(file, 'rb') as fo:
---> 29         dict = pickle.load(fo)
     30         dict = dict.encode('ascii', 'ignore')
     31     return dict

UnicodeDecodeError: 'ascii' codec can't decode byte 0x8b in position 6: ordinal not in range(128)

我不知道该怎么办，我已经尝试了所有方法，但仍然遇到相同的错误。这是我的代码:

# IMAGE RECOGNITION

# Tensorflow and numpy to create the neural network
import tensorflow as tf
import numpy as np

# Matplotlib to plot info to show our results
import matplotlib.pyplot as plt

# OS to load files and save checkpoints
import os

# LOADING THE DATA:

# LOADING CIFAR data from file:
# Load cifar data from file
# Load MNIST data from tf examples

# Load cifar data from file

image_height = 32
image_width = 32

color_channels = 3

model_name = "cifar"


def unpickle(file):
    import pickle
    with open(file, 'rb') as fo:
        dict = pickle.load(fo)

    return dict




train_data = np.array([])
train_labels = np.array([])

# Load all the data batches.
for i in range(5):
    data_batch = unpickle( 'data_batch_' + str(i + 1))

    train_data = np.append(train_data, data_batch[b'data'])
    train_labels = np.append(train_labels, data_batch[b'labels'])

# Load the eval batch.
eval_batch = unpickle( 'test_batch')

eval_data = eval_batch[b'data']
eval_labels = eval_batch[b'labels']

# Load the english category names.
category_names_bytes = unpickle('batches.meta')[b'label_names']
category_names = list(map(lambda x: x.decode("utf-8"), category_names_bytes))


# TODO: Process Cifar data

def process_data(data):
    float_data = np.array(data, dtype=float) / 255.0

    reshaped_data = np.reshape(float_data, (-1, color_channels, image_height, image_width))

    # The incorrect image

    transposed_data = np.transpose(reshaped_data, [0, 2, 3, 1])

    return transposed_data


train_data = process_data(train_data)

eval_data = process_data(eval_data)

# TODO: The Neural Network
# CONVOLUTIONAL NEURAL NETWORK CLASS:

谢谢!

Answer 1

尝试:

pickle.load(fo, encoding='latin1')

这可能是 Python 2/3 兼容性问题。顺便说一句，你应该尽量不要使用像 dict 这样的东西作为变量名，因为它会覆盖 Python 内置函数。