機器學習中最經典的例子就是MNIST通過圖片來識別0~9的數字，這篇文章將介紹如何將基於Tensorflow MNIST整合到Diego1#機器人中作為一個節點，此節點將訂閱Image訊息，通過MNIST識別後將結果釋出訊息給訊飛語音節點，訊飛語音節點會告訴我們識別的數字是幾。

相關原始碼已經上傳到本人的github。

1. 安裝Tensorflow

只需一句命令即可安裝

pip install tensorflow

官方有4中安裝方法，在這裡選擇直接安裝的方式

2. 建立diego_tensorflow_mnist 包

catkin_create_pkg diego_tensorflow_mnist std_msgs rospy roscpp cv_bridge
 

在diego_tensorflow_mnist目錄下建立scripts和launch目錄

scripts目錄用於存放python的原始碼
launch目錄用於存放ROS launch檔案

下載相關程式碼到scripts目錄

3.ROS節點

有關nnist的演算法都已經寫好，我們只需要呼叫其中的功能封裝成ROS節點即可，有關封裝的程式碼請見tensorflow_in_ros_mnist.py

#!/usr/bin/env python

import rospy
from sensor_msgs.msg import Image
from std_msgs.msg import
 
 Int16
from std_msgs.msg import String
from cv_bridge import CvBridge
import cv2
import numpy as np
import tensorflow as tf


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

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

def conv2d(x, W):
  return tf.nn.conv2d(x, W, strides=[1, 1, 1, 1], 
                      padding='SAME')

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


def makeCNN(x,keep_prob):
    # --- define CNN model
    W_conv1 = weight_variable([5, 5, 1, 32])
    b_conv1 = bias_variable([32])
    h_conv1 = tf.nn.relu(conv2d(x, W_conv1) + b_conv1)

    h_pool1 = max_pool_2x2(h_conv1)

    W_conv2 = weight_variable([3, 3, 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)

    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_conv = tf.nn.softmax(tf.matmul(h_fc1_drop, W_fc2) + b_fc2)

    return y_conv



class RosTensorFlow():
    def __init__(self):

    rospy.init_node('rostensorflow')

    # Set the shutdown function (stop the robot)
        rospy.on_shutdown(self.shutdown)

        model_path = rospy.get_param("~model_path", "")
        image_topic = rospy.get_param("~image_topic", "")

        self._cv_bridge = CvBridge()

        self.x = tf.placeholder(tf.float32, [None,28,28,1], name="x")
        self.keep_prob = tf.placeholder("float")
        self.y_conv = makeCNN(self.x,self.keep_prob)

        self._saver = tf.train.Saver()
        self._session = tf.InteractiveSession()

        init_op = tf.global_variables_initializer()
        self._session.run(init_op)

        self._saver.restore(self._session, model_path+"/model.ckpt")

        self._sub = rospy.Subscriber(image_topic, Image, self.callback, queue_size=1)
        #self._pub = rospy.Publisher('result', Int16, queue_size=1)
        self._pub = rospy.Publisher('xfwords', String, queue_size=1)


    def callback(self, image_msg):
        cv_image = self._cv_bridge.imgmsg_to_cv2(image_msg, "bgr8")
        cv_image_gray = cv2.cvtColor(cv_image, cv2.COLOR_RGB2GRAY)
        ret,cv_image_binary = cv2.threshold(cv_image_gray,128,255,cv2.THRESH_BINARY_INV)
        cv_image_28 = cv2.resize(cv_image_binary,(28,28))
        np_image = np.reshape(cv_image_28,(1,28,28,1))
        predict_num = self._session.run(self.y_conv, feed_dict={self.x:np_image,self.keep_prob:1.0})
        answer = np.argmax(predict_num,1)
        rospy.loginfo('%d' % answer)
        self._pub.publish(str(answer))
        rospy.sleep(3) 

    def shutdown(self):
        rospy.loginfo("Stopping the tensorflow nnist...")
        rospy.sleep(1)   

if __name__ == '__main__':
    try:
        RosTensorFlow()
        rospy.spin()
    except rospy.ROSInterruptException:
        rospy.loginfo("RosTensorFlow has started.")

有關MNIST具體演算法實現部分網上有很多教程，這裡只說明與ROS整合部分

class RosTensorFlow():
    def __init__(self):

    rospy.init_node('rostensorflow')

    # Set the shutdown function (stop the robot)
        rospy.on_shutdown(self.shutdown)

        model_path = rospy.get_param("~model_path", "")
        image_topic = rospy.get_param("~image_topic", "")

在RosTensorFlow類的開始部分，是標準的節點定義方法，model_path變數用於獲取launch檔案中定義的model的路徑，image_topic變數用於獲取launch檔案中定義image主題

        self._sub = rospy.Subscriber(image_topic, Image, self.callback, queue_size=1)

以上這段程式碼是讓該節點訂閱image topic，並且知道回撥函式

        self._pub = rospy.Publisher('xfwords', String, queue_size=1)

以上這段程式碼定義將釋出訊飛語音主題，xfwords

    def callback(self, image_msg):
        cv_image = self._cv_bridge.imgmsg_to_cv2(image_msg, "bgr8")
        cv_image_gray = cv2.cvtColor(cv_image, cv2.COLOR_RGB2GRAY)
        ret,cv_image_binary = cv2.threshold(cv_image_gray,128,255,cv2.THRESH_BINARY_INV)
        cv_image_28 = cv2.resize(cv_image_binary,(28,28))
        np_image = np.reshape(cv_image_28,(1,28,28,1))
        predict_num = self._session.run(self.y_conv, feed_dict={self.x:np_image,self.keep_prob:1.0})
        answer = np.argmax(predict_num,1)
        rospy.loginfo('%d' % answer)
        self._pub.publish(str(answer))
        rospy.sleep(3) 

主要的處理都在callback回撥函式中，首先將從image主題中經過一系列的處理轉換成numpy陣列，然後呼叫tensorflow進行識別，將可能的結果過放在predict_num陣列中，取其中最有可能的值，就是結果，將結果作為訊飛語音topic傳送出去

4.launch檔案

在launch資料夾下建立一個名為nnist.launch的檔案，檔案內容如下：

<launch>

    <node pkg="diego_tensorflow_nnist" name="tensorflow_in_ros_mnist" type="tensorflow_in_ros_mnist.py" output="screen">
        <param name="image_topic" value="/usb_cam/image_raw" />
        <param name="model_path" value="$(find diego_tensorflow_nnist)/scripts/model" />
    </node> 
</launch>

相關的主題，和路徑可以在這裡修改

5.啟動節點

roscore
rosrun xfei_asr tts_subscribe_speak
roslaunch usb_cam usb_cam-test.launch
roslaunch diego_tensorflow_nnist nest.launch

啟動後我們就可以在紙上面寫幾個數字，放在攝像頭前，diego1#會告訴你數字是多少。