TensorFlow与Flask结合打造手写体数字识别

• flask
• Tensorflow

定义模型model

mnist_testdemo/mnist/model.py

线性模型

import tensorflow as tf


# Y=W*x+b 线性模型
def regression(x):
    W = tf.Variable(tf.zeros([784, 10]), name="W")
    b = tf.Variable(tf.zeros([10]), name="b")
    y = tf.nn.softmax(tf.matmul(x, W) + b)

    return y, [W, b]

卷积模型 

# 卷积模型
def convolutional(x, keep_prob):
    # 卷积层
    def conv2d(x, W):
        return tf.nn.conv2d(x, W, [1, 1, 1, 1], padding='SAME')

    # 池化层
    def max_pool_2x2(x):
        return tf.nn.max_pool(x, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding='SAME')

    # 定义权重
    def weight_variable(shape):
        initial = tf.truncated_normal(shape, stddev=0.1)
        return tf.Variable(initial)

    # 偏置
    def bias_variable(shape):
        initial = tf.constant(0.1, shape=shape)
        return tf.Variable(initial)

    x_image = tf.reshape(x, [-1, 28, 28, 1])
    W_conv1 = weight_variable([5, 5, 1, 32])
    b_conv1 = bias_variable([32])
    h_conv1 = tf.nn.relu(conv2d(x_image, W_conv1) + b_conv1)
    h_pool1 = max_pool_2x2(h_conv1)

    W_conv2 = weight_variable([5, 5, 32, 64])
    b_conv2 = bias_variable([64])
    h_conv2 = tf.nn.relu(conv2d(h_pool1, W_conv2) + b_conv2)
    h_pool2 = max_pool_2x2(h_conv2)

    # full connection
    W_fc1 = weight_variable([7 * 7 * 64, 1024])
    b_fc1 = bias_variable([1024])
    h_pool2_flat = tf.reshape(h_pool2, [-1, 7 * 7 * 64])
    h_fc1 = tf.nn.relu(tf.matmul(h_pool2_flat, W_fc1) + b_fc1)

    h_fc1_drop = tf.nn.dropout(h_fc1, keep_prob)

    W_fc2 = weight_variable([1024, 10])
    b_fc2 = bias_variable([10])
    y = tf.nn.softmax(tf.matmul(h_fc1_drop, W_fc2) + b_fc2)

    return y, [W_conv1, b_conv1, W_conv2, b_conv2, W_fc1, b_fc1, W_fc2, b_fc2]

定义数据

mnist_testdemo/mnist/input_data.py 

from __future__ import absolute_import
from __future__ import division
from __future__ import print_function

import gzip
import os
import tempfile

import numpy
from six.moves import urllib
from six.moves import xrange
import tensorflow as tf
from tensorflow.contrib.learn.python.learn.datasets.mnist import read_data_sets

训练模型 

训练线性模型 

mnist_testdemo/mnist/regression.py

import os

import input_data
import model
import tensorflow as tf

# 从input_data中下载数据到MNIST_data
data = input_data.read_data_sets('MNIST_data', one_hot=True)

# create model
with tf.variable_scope("regression"):
    # 用户输入占位符
    x = tf.placeholder(tf.float32, [None, 784])
    y, variables = model.regression(x)

# train
y_ = tf.placeholder("float", [None, 10])
cross_entropy = -tf.reduce_sum(y_ * tf.log(y))
# 训练步骤
train_step = tf.train.GradientDescentOptimizer(0.01).minimize(cross_entropy)
# 预测
correct_prediction = tf.equal(tf.argmax(y, 1), tf.argmax(y_, 1))
# 准确度
accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))

# 保存训练变量参数
saver = tf.train.Saver(variables)
# 开始训练
with tf.Session() as sess:
    sess.run(tf.global_variables_initializer())
    for _ in range(20000):
        batch_xs, batch_ys = data.train.next_batch(100)
        sess.run(train_step, feed_dict={x: batch_xs, y_: batch_ys})

    # 打印测试集和训练集的精准度
    print((sess.run(accuracy, feed_dict={x:data.test.images, y_:data.test.labels})))

    # 保存训练好的模型
    path = saver.save(
        sess,os.path.join(os.path.dirname(__file__),'data','regression.ckpt'),
        write_meta_graph=False,write_state=False)
    print("Saved:", path)

生成mnist_testdemo/mnist/data/regression.ckpt.data-00000-of-00001和mnist_testdemo/mnist/data/regression.ckpt.index 

训练卷积模型 

mnist_testdemo/mnist/convolutional.py 

import os
import model
import tensorflow as tf
import input_data

data = input_data.read_data_sets('MNIST_data', one_hot=True)

#model
with tf.variable_scope("convolutional"):
    x = tf.placeholder(tf.float32, [None, 784], name='x')
    keep_prob = tf.placeholder(tf.float32)
    y, variables = model.convolutional(x, keep_prob)

#train
y_ = tf.placeholder(tf.float32, [None, 10], name='y')
cross_entropy = -tf.reduce_sum(y_ * tf.log(y))
# 随机梯度下降
train_step = tf.train.AdamOptimizer(1e-4).minimize(cross_entropy)
correct_prediction = tf.equal(tf.argmax(y, 1), tf.argmax(y_, 1))
accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))

saver = tf.train.Saver(variables)

with tf.Session() as sess:
    merged_summary_op = tf.summary.merge_all()
    summay_writer = tf.summary.FileWriter('./mnist_log/1', sess.graph)
    summay_writer.add_graph(sess.graph)
    sess.run(tf.global_variables_initializer())

    for i in range(20000):
        batch = data.train.next_batch(50)
        if i % 100 == 0:
            train_accuracy = accuracy.eval(feed_dict={x: batch[0], y_: batch[1], keep_prob: 1.0})
            print("step %d, training accuracy %g" % (i, train_accuracy))
        sess.run(train_step, feed_dict={x: batch[0], y_: batch[1], keep_prob: 0.5})

    print(sess.run(accuracy, feed_dict={x: data.test.images, y_: data.test.labels, keep_prob: 1.0}))

    path = saver.save(
        sess, os.path.join(os.path.dirname(__file__), 'data', 'convalutional.ckpt'),
        write_meta_graph=False, write_state=False)

    print("Saved:", path)

生成 mnist_testdemo/mnist/data/convalutional.ckpt.data-00000-of-00001和mnist_testdemo/mnist/data/convalutional.ckpt.index

集成flask 

mnist_testdemo/main.py

# -*- coding:utf-8 -*-
import numpy as np
import tensorflow as tf
from flask import Flask, jsonify, render_template, request
import pprint

from mnist import model

x = tf.placeholder("float", [None, 784])
sess = tf.Session()

# 取出训练好的线性模型
with tf.variable_scope("regression"):
    y1, variables = model.regression(x)

saver = tf.train.Saver(variables)
saver.restore(sess, "mnist/data/regression.ckpt")

# 取出训练好的卷积模型
with tf.variable_scope("convolutional"):
    keep_prob = tf.placeholder("float")
    y2, variables = model.convolutional(x, keep_prob)

saver = tf.train.Saver(variables)
saver.restore(sess, "mnist/data/convalutional.ckpt")


# 根据输入调用线性模型并返回识别结果
def regression(input):
    return sess.run(y1, feed_dict={x: input}).flatten().tolist()


# 根据输入调用卷积模型并返回识别结果
def convolutional(input):
    return sess.run(y2, feed_dict={x: input, keep_prob: 1.0}).flatten().tolist()


app = Flask(__name__)


@app.route('/api/mnist', methods=['POST'])
def mnist():
    # pprint.pprint(request.json)
    input = ((255 - np.array(request.json, dtype=np.uint8)) / 255.0).reshape(1, 784)
    output1 = regression(input)
    output2 = convolutional(input)
    pprint.pprint(output1)
    pprint.pprint(output2)
    return jsonify(results=[output1, output2])


@app.route('/')
def main():
    return render_template('index.html')


if __name__ == '__main__':
    app.debug = True
    app.run(host='0.0.0.0', port=8889)

 js核心代码

drawInput() {
        var ctx = this.input.getContext('2d');
        var img = new Image();
        img.onload = () => {
            var inputs = [];
            var small = document.createElement('canvas').getContext('2d');
            small.drawImage(img, 0, 0, img.width, img.height, 0, 0, 28, 28);
            var data = small.getImageData(0, 0, 28, 28).data;
            for (var i = 0; i < 28; i++) {
                for (var j = 0; j < 28; j++) {
                    var n = 4 * (i * 28 + j);
                    inputs[i * 28 + j] = (data[n + 0] + data[n + 1] + data[n + 2]) / 3;
                    ctx.fillStyle = 'rgb(' + [data[n + 0], data[n + 1], data[n + 2]].join(',') + ')';
                    ctx.fillRect(j * 5, i * 5, 5, 5);
                }
            }
            if (Math.min(...inputs) === 255) {
                return;
            }
            $.ajax({
                url: '/api/mnist',
                type: 'POST',
                contentType: 'application/json',
                data: JSON.stringify(inputs),
                success: (data) => {
                    data = JSON.parse(data);
                    for (let i = 0; i < 2; i++) {
                        var max = 0;
                        var max_index = 0;
                        for (let j = 0; j < 10; j++) {

                            var value = Math.round(data.results[i][j] * 1000);
                            if (value > max) {
                                max = value;
                                max_index = j;
                            }
                            var digits = String(value).length;
                            for (var k = 0; k < 3 - digits; k++) {
                                value = '0' + value;
                            }
                            var text = '0.' + value;
                            if (value > 999) {
                                text = '1.000';
                            }
                            $('#output tr').eq(j + 1).find('td').eq(i).text(text);
                        }
                        for (let j = 0; j < 10; j++) {
                            if (j === max_index) {
                                $('#output tr').eq(j + 1).find('td').eq(i).addClass('success');
                            } else {
                                $('#output tr').eq(j + 1).find('td').eq(i).removeClass('success');
                            }
                        }
                    }
                }
            });
        };
        img.src = this.canvas.toDataURL();
    }

• 前端将数据inputs以json传入/api/mnist
• regression(input)和convolutional(input)调用模型feed_dict喂参数返回结果
• 源码：链接：https://pan.baidu.com/s/17omtWb5EsG7pvpZu_b6WaA 
  提取码：vl5m 
   

参考：https://blog.csdn.net/welggy/article/details/100154164

 

训练运行查看CPU性能

可以看到使用CPU训练利用率很高，而且训练速度也是超级慢

使用GPU训练的话相对来讲就快多了