基于tensorflow框架的手写数字识别+基于pyqt5的GUI界面+opencv图片数字切割.zip

''' Created on 2018年8月8日 @author: Freedom ''' from PyQt5.Qt import QWidget, QColor, QPixmap, QIcon, QSize, QCheckBox from PyQt5.QtWidgets import QHBoxLayout, QVBoxLayout, QPushButton, QSplitter,\ QComboBox, QLabel, QSpinBox, QFileDialog from PaintBoard import PaintBoard from PyQt5.QtCore import Qt from PyQt5.QtGui import QFont, QPalette, QBrush import tensorflow as tf from PIL import ImageGrab, Image from PyQt5.QtWidgets import QApplication import PIL.ImageOps import cv2 import numpy as np import os class MainWidget(QWidget): def __init__(self, Parent=None): ''' Constructor ''' self.b_list = list() self.num = 0 self.pix = [] self.result = 0 super().__init__(Parent) self.__InitData() #先初始化数据，再初始化界面 self.__InitView() self.setWindowFlags(Qt.FramelessWindowHint) # 窗体无边框 self.initUI() print(self.width,self.height) def __InitData(self): ''' 初始化成员变量 ''' self.__paintBoard = PaintBoard(self) def __InitView(self): ''' 初始化界面 ''' self.setFixedSize(480,300) self.setWindowTitle("手写数字识别") palette = QPalette() palette.setBrush(QPalette.Background, QBrush(QPixmap("123.jpg"))) self.setPalette(palette) main_layout = QHBoxLayout(self) #新建一个水平布局作为本窗体的主布局 main_layout.setSpacing(10) #设置主布局内边距以及控件间距为10px main_layout.addWidget(self.__paintBoard) #在主界面左侧放置画板 sub_layout = QVBoxLayout() #新建垂直子布局用于放置按键 sub_layout.setContentsMargins(15, 15, 10, 10) #设置此子布局和内部控件的间距 self.__btn_Clear = QPushButton("清空画板") self.__btn_Clear.setFixedSize(116,25) self.__btn_Clear.setStyleSheet("QPushButton{color:rgb(101,153,26)}" "QPushButton{background-color: #EED5D2}" "QPushButton:hover{color:red}" "QPushButton{border-radius:6px}" "QPushButton:pressed{background-color:rgb(180,180,180);border: None;}") self.__btn_Clear.setParent(self) #设置父对象为本界面 self.__btn_Clear.clicked.connect(self.Clear) #将按键按下信号与画板清空函数相关联 sub_layout.addWidget(self.__btn_Clear) self.__btn_Save = QPushButton("图片识别") self.__btn_Save.setFixedSize(116,25) self.__btn_Save.setStyleSheet("QPushButton{color:rgb(101,153,26)}" "QPushButton{background-color:#FFE1FF}" "QPushButton:hover{color:red}" "QPushButton{border-radius:6px}" "QPushButton:pressed{background-color:rgb(180,180,180);border: None;}") self.__btn_Save.setParent(self) self.__btn_Save.clicked.connect(self.btn_recognize_on_clicked) sub_layout.addWidget(self.__btn_Save) self.__btn_Add = QPushButton("添加图片并识别") self.__btn_Add.setFixedSize(116,25) self.__btn_Add.setStyleSheet("QPushButton{color:rgb(101,153,26)}" "QPushButton{background-color:#FFE1EE}" "QPushButton:hover{color:red}" "QPushButton{border-radius:6px}" "QPushButton:pressed{background-color:rgb(180,180,180);border: None;}") self.__btn_Add.setParent(self) self.__btn_Add.clicked.connect(self.add_pic) sub_layout.addWidget(self.__btn_Add) self.__btn_Quit = QPushButton("程序退出") self.__btn_Quit.setFixedSize(116,25) self.__btn_Quit.setStyleSheet("QPushButton{color:rgb(101,153,26)}" "QPushButton{background-color:#CFCFCF}" "QPushButton:hover{color:red}" "QPushButton{border-radius:6px}" "QPushButton:pressed{background-color:rgb(180,180,180);border: None;}") self.__btn_Quit.setParent(self) #设置父对象为本界面 self.__btn_Quit.clicked.connect(self.Quit) sub_layout.addWidget(self.__btn_Quit) # splitter = QSplitter(self) #占位符 # sub_layout.addWidget(splitter) self.__label_penThickness = QLabel(self) self.__label_penThickness.setText("识别结果：") self.__label_penThickness.setFont(QFont("Roman times",8,QFont.Bold)) self.__label_penThickness.setFixedHeight(18) sub_layout.addWidget(self.__label_penThickness) self.label_result = QLabel(self) self.label_result.setStyleSheet("QLabel{border:1px solid #EED5D2;border-radius:4px;color:#FFE4C4;font-size:25px;font-weight:bold;text-align: center;}") self.label_result.setFixedSize(140,28) sub_layout.addWidget(self.label_result) main_layout.addLayout(sub_layout) #将子布局加入主布局 def btn_recognize_on_clicked(self): if self.__paintBoard.pos_xy == []: self.label_result.setText('') # 显示识别结果 return self.update() bbox = ( self.width//2-226,self.height//2-155, self.width//2+53,self.height//2+124) print(bbox) im = ImageGrab.grab(bbox) # 截屏，手写数字部分 im = im.resize((28, 28), Image.ANTIALIAS) # 将截图转换成 28 * 28 像素 recognize_result,flag = self.recognize_img(im) # 调用识别函数 if flag == 1: return cnn_acc,knn_acc,tree_acc = self.acc() if cnn_acc<knn_acc: cnn_acc,knn_acc=knn_acc,cnn_acc if recognize_result != 0: #im.save('./0/mnist_0_%d.png' %self.num) self.num += 1 self.label_result.setText(str(recognize_result)) # 显示识别结果 self.label_cnn_acc.setText(str("cnn:%.6f"%cnn_acc)) self.label_knn_acc.setText(str("knn:%.6f"%knn_acc)) self.label_tree_acc.setText(str("DTree:%.6f"%tree_acc)) self.update() def recognize_img(self, img): # 手写体识别函数 myimage = img.convert('L') # 转换成灰度图 tv = list(myimage.getdata()) # 获取图片像素值 tva = [(255 - x) * 1.0 / 255.0 for x in tv] # 转换像素范围到[0 1], 0是纯白 1是纯黑 flag = 0 if self.pix == tva: flag = 1 return self.result,flag self.pix = tva init = tf.global_variables_initializer() saver = tf.train.Saver with tf.Session() as sess: sess.run(init) saver = tf.train.import_meta_graph("./model_version4/model.ckpt.meta") # 载入模型结构 saver.restore(sess, "./model_version4/model.ckpt") # 载入模型参数 graph = tf.get_default_graph() # 加载计算图 x = graph.get_tensor_by_name("x:0") # 从模型中读取占位符变量 keep_prob = graph.get_tensor_by_name("keep_prob:0") y_conv = graph.get_tensor_by_name("y_conv:0") # 关键的一句 从模型中读取占位符变量 prediction = tf.argmax(y_conv, 1) predint = prediction.eval(feed_dict={x: [tva], keep_prob: 1.0}, session=sess) # feed_dict输入数据给placeholder占位符 self.result = predint[0] print(predint[0]) return predint[0],flag def Clear(self): #清空画板 self.__paintBoard._InitData() self.__paintBoard.pos_xy = [] self.l