利用python qt和机器学习实现医学影像X光片的识别与分析.zip

from PySide2.QtWidgets import QApplication, QMainWindow, QPushButton, QPlainTextEdit, QMessageBox, QFileDialog, \ QTextBrowser, QLabel from PySide2.QtUiTools import QUiLoader from PySide2 import QtGui from PySide2.QtCore import Signal, QObject,QCoreApplication from PIL import Image import torch.utils.data from torchvision.utils import save_image import os import os os.environ["CUDA_VISIBLE_DEVICES"] = "-1" import torch import torch.nn as nn import torch.nn.functional as F import cv2 import numpy as np from Module.随机生成数字 import generator import Module.识别数字 as RegconizeNum # 模块从Module包里导入“识别数字” device = torch.device("cuda" if torch.cuda.is_available() else "cpu") # 没gpu的话就用cpu class MySignals(QObject): # 定义一种信号，参赛是str，即文件的地址 ms = Signal(str) global_ms = MySignals() # 实例化信号 input_num_ms = MySignals() class ImgWindow(): # 显示图片的窗口 def __init__(self): super().__init__() # 使用ui文件导入定义界面类 self.ui = QUiLoader().load('img_window.ui') self.ui.Button_exit.clicked.connect(self.exit_b) # global_ms.ms.connect(self.load_img) # 连接信号与槽 def exit_b(self): os.remove("temp.png") # 删除生成的临时文件 self.ui.close() def load_img(self, object): im = Image.open(object) # 这里把原来的jpg转化成png之后打开 im.save('temp.png') pixmap = QtGui.QPixmap('temp.png') label = self.ui.img_label label.setPixmap(pixmap) # 加载图片 label.setScaledContents(True) # 自适应 class InputNumWindow(): # 用户输入图片张数的窗口 def __init__(self): self.ui = QUiLoader().load('input_num.ui') self.ui.ok_btn.clicked.connect(self.get_num) self.ui.cancel_btn.clicked.connect(self.close_ui) def get_num(self): num = self.ui.user_input_num.text() input_num_ms.ms.emit(num) self.close_ui() def close_ui(self): self.ui.close() class MainWindow(): # 主窗口 def __init__(self): super().__init__() # 使用ui文件导入定义界面类 self.ui = QUiLoader().load('my_ui.ui') self.ui.Button_loadmodel.clicked.connect(self.load_model) self.ui.Button_openimg.clicked.connect(self.open_img) self.ui.Button_randnum.clicked.connect(self.open_input_window) self.ui.Button_consequence.clicked.connect(self.predict_res) input_num_ms.ms.connect(self.get_mynum) def get_mynum(self,num): self.my_num=num self.input_randnum() def load_model(self): FileDialog = QFileDialog(self.ui.Button_loadmodel) # 实例化 FileDialog.setFileMode(QFileDialog.AnyFile) # 可以打开任何文件 model_file, _ = FileDialog.getOpenFileName(self.ui.Button_loadmodel, 'open file', './', 'model files (*.pth)') # 改变Text里面的文字 self.ui.View_model_log.setPlainText("成功加载模型\n模型路径:" + model_file) def open_img(self): # 这里和load_model差不多 FileDialog = QFileDialog(self.ui.Button_openimg) FileDialog.setFileMode(QFileDialog.AnyFile) image_file, _ = FileDialog.getOpenFileName(self.ui.Button_openimg, 'open file', './Handwriting num pic', 'Image files (*.jpg *.gif *.png *.jpeg)') if not image_file: QMessageBox.warning(self.ui.Button_openimg, "警告", "文件错误或打开文件失败！", QMessageBox.Yes) return self.ui.View_img_log.setPlainText("成功加载图片\n图片路径:" + image_file) self.window2 = ImgWindow() global_ms.ms.emit(image_file) # 注意只有先实例化之后 发送信号 对应的槽才会执行 self.window2.ui.show() def open_input_window(self): self.window3 = InputNumWindow() self.window3.ui.show() def input_randnum(self):#num_ms是通过messege接受到的图片个数信息 num = int(self.my_num) G = generator(100, 3136).to(device) model_file = self.ui.View_model_log.toPlainText().split('路径:')[1] # 模型保存地址 my_net = torch.load(model_file, map_location=torch.device('cpu')) # 加载模型,没gpu的话将内存定位cpu G.load_state_dict(my_net) # num = 10 # 用户输入需要的图像张数 filename = "Handwriting num pic" current_path = os.getcwd() # 返回当前 path_item = os.listdir(current_path) # 返回（列表）将当前目录的所有内容 picfile_path = "{}\Handwriting num pic".format(current_path) # 图片保存进哪个文件夹的路径 if filename not in path_item: os.mkdir(filename) # 在当前目录创建文件夹 for i in range(num): z = torch.randn(1, 100).to(device) # z = torch.ones(100).unsqueeze(0).to(device) # z = Variable(torch.randn(1, 100)) fake_img = G(z) path = "./{}/pic_{}.png".format(filename, i + 1) # 保存图片吗的路径 save_image(fake_img, path.format(i)) str = "成功生成{num}张手写数字图\n图片路径:{path}".format(num=num, path=picfile_path) self.ui.View_randnum_log.setPlainText(str) def predict_res(self): image_file = self.ui.View_img_log.toPlainText().split('路径:')[1] img = cv2.imread(image_file, 0) # 加载图像,RGB是3通道图像0表示以一通道的灰度图表示 img = cv2.resize(img, (28, 28)) # 缩放到28*28 img = RegconizeNum.nor(img) # 归一化到0-1 DEVICE = torch.device("cuda" if torch.cuda.is_available() else "cpu") # 让torch判断是否使用GPU，建议使用GPU环境，因为会快很多 img = torch.from_numpy(img).unsqueeze(0).unsqueeze(0).float().to(DEVICE) # 图像转为tensor格式 model = RegconizeNum.ConvNet().to(DEVICE) model_file = self.ui.View_model_log.toPlainText().split('路径:')[1] # 模型保存地址 my_net = torch.load(model_file) model.load_state_dict(my_net) model.eval() output = model(img) # 预测 pred = output.max(1, keepdim=True)[1] self.ui.View_predict_log.setPlainText("预测识别的结果为：" + str(pred[0][0].data.cpu().numpy())) app = QApplication([]) start = MainWindow() start.ui.show() app.exec_()