在入门之前,我们需要开发工具,本文使用 JupyterLab,可以用 conda 或者 pip 方式安装。
// conda 方式
conda install -c conda-forge jupyterlab
// or pip 方式
pip install jupyterlab
conda 源更新比较缓慢,推荐还是用 pip。
启用:
jupyter-lab
为了在不同的 conda 虚拟环境下使用 jupyterlab,可以安装插件 nb_conda_kernels
。
conda install -n tf2 nb_conda_kernels
下面就可以运行一个 hello world
和开发 Tensorflow了。
引用
import matplotlib.pyplot as plt
from typing import Dict, Text
import numpy as np
import tensorflow as tf
import tensorflow_datasets as tfds
import tensorflow_recommenders as tfrs
import os
import ssl
os.environ['HTTP_PROXY'] = 'http://0.0.0.0:8888'
os.environ['HTTPS_PROXY'] = 'http://0.0.0.0:8888'
ssl._create_default_https_context = ssl._create_unverified_context
下载 MNIST 数据集
# MNIST data.
mnist_train = tfds.load(name="mnist", split="train", data_dir = os.path.join(os.getcwd(), "data"))
效果:
<PrefetchDataset shapes: {image: (28, 28, 1), label: ()}, types: {image: tf.uint8, label: tf.int64}>
图片格式主要是 28*28,我们可以写个代码将数据集保存为图片,看看图片效果。
转为图片
for mnist_example in mnist_train.take(1): # 只取一个样本
image, label = mnist_example["image"], mnist_example["label"]
plt.imshow(image.numpy()[:, :, 0].astype(np.float32), cmap=plt.get_cmap("gray"))
print("Label: %d" % label.numpy())
说明数据我们已经拿到手,有了数据,我们可以开始往下进行。
获取训练集和测试集
(x_train, y_train), (x_test, y_test) = tf.keras.datasets.mnist.load_data(
path = os.path.join(os.getcwd(), "data/mnist.npz")
)
初始化和灰度化
统一图片大小和灰度化:
x_train = x_train.reshape(x_train.shape[0], 28, 28, 1)
x_test = x_test.reshape(x_test.shape[0], 28, 28, 1)
input_shape = (28, 28, 1)
x_train = x_train.astype('float32')
x_test = x_test.astype('float32')
x_train /= 255
x_test /= 255
print('x_train shape: ', x_train.shape)
print('Number of images in x_train', x_train.shape[0])
print('Number of images in x_test', x_test.shape[0])
建立自然网络模型
# Importing the required Keras modules containing model and layers
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Dense, Conv2D, Dropout, Flatten, MaxPooling2D
# Creating a Sequential Model and adding the layers
model = Sequential()
model.add(Conv2D(28, kernel_size=(3,3), input_shape=input_shape))
model.add(MaxPooling2D(pool_size=(2, 2)))
model.add(Flatten()) # Flattening the 2D arrays for fully connected layers
model.add(Dense(128, activation=tf.nn.relu))
model.add(Dropout(0.2))
model.add(Dense(10,activation=tf.nn.softmax))
编译模型
model.compile(optimizer='adam',
loss='sparse_categorical_crossentropy',
metrics=['accuracy'])
model.fit(x=x_train,y=y_train, epochs=10)
model.evaluate(x_test, y_test)
测试
image_index = 5555
plt.imshow(x_test[image_index].reshape(28, 28),cmap='Greys')
pred = model.predict(x_test[image_index].reshape(1, 28, 28, 1))
print(pred.argmax())
image_index = 6666
plt.imshow(x_test[image_index].reshape(28, 28),cmap='Greys')
pred = model.predict(x_test[image_index].reshape(1, 28, 28, 1))
print(pred.argmax())
总结
初步学习使用 MNIST 数据集做训练和对手写数字的识别测试,开启 tensorflow 的入门。
THE MNIST DATABASE of handwritten digits[1]
Image Classification in 10 Minutes with MNIST Dataset[2]
参考
[1] THE MNIST DATABASE of handwritten digits http://yann.lecun.com/exdb/mnist/
[2] Image Classification in 10 Minutes with MNIST Dataset https://towardsdatascience.com/image-classification-in-10-minutes-with-mnist-dataset-54c35b77a38d