tensorflow实现图像分类识别

import tensorflow as tf
from tensorflow import keras
import numpy as np
import matplotlib.pyplot as plt

# print(tf.__version__)
fashion_mnsit=keras.datasets.fashion_mnist
# print(type(fashion_mnsit))
# print(fashion_mnsit.load_data)
#train_images为训练集，test_images为测试集
'''
对抗神经对抗网络亦是采用训练集测试集对抗训练
'''
(train_images, train_labels), (test_images, test_labels) = fashion_mnsit.load_data()
# print(train_images[1])
# #检查数据
# plt.figure()
# plt.imshow(train_images[1])
# plt.colorbar()
# plt.grid(False)
# plt.show()
#将train_images, test_images值除以255，方便传入神经网络层。

train_images = train_images / 255.0
# print('train_images的类型为:'+str(type(train_images)))

test_images = test_images / 255.0
#设置层
model = keras.Sequential([
#第一层将数据传输设置成 28*28 的一维数组。格式化
    keras.layers.Flatten(input_shape=(28, 28)),
#设置神经层，每层128个神经元
    keras.layers.Dense(128, activation='relu'),
#分10组打分。
    keras.layers.Dense(10)
])
model.compile(
    #梯度下降算法
              optimizer='adam',
              loss=tf.keras.losses.SparseCategoricalCrossentropy(from_logits=True),
              metrics=None,
              loss_weights=None,
              weighted_metrics=None,
              run_eagerly=None,
              steps_per_execution=None,

)
#向神经网络馈送数据模型
model.fit(train_images,train_labels,epochs=10)
# #评估损失
test_acc = model.evaluate(test_images,test_labels,verbose=2)
print('准确率为：'+str(test_acc))