PyQt训练BP模型时，显示waiting动图（多线程）

1、实现效果

2、相关代码

实现BP训练模型的线程类

 class WorkThread(QtCore.QThread):
     finish_trigger = QtCore.pyqtSignal()  # 关闭waiting_gif
     result_trigger = QtCore.pyqtSignal(pd.Series)  # 传递预测结果信号
     evaluate_trigger = QtCore.pyqtSignal(list)  # 传递正确率信号
 
     def __int__(self):
         super(WorkThread, self).__init__()
 
     def init(self, dataset, feature, label, info):
         self.dataset = dataset
         self.feature = feature
         self.label = label
         self.info = info
 
     # 可以认为，run()函数就是新的线程需要执行的代码
     def run(self):
         self.BP()
 
     def BP(self):
         """
         BP神经网络，返回标签的预测数据
         :param parent:
         :param dataset:
         :param feature:
         :param label:
         :param info:
         :return:
         """
         dataset = self.dataset
         feature = self.feature
         label = self.label
         info = self.info
 
         input_dim = len(feature)
         data_x = dataset[feature]  # 特征数据
         data_y = dataset[label]  # 标签数据
 
         x_train, x_test, y_train, y_test = train_test_split(data_x, data_y, test_size=info[0][3])
 
         # **********************建立一个简单BP神经网络模型*********************************
         self.model = Sequential()  # 声明一个顺序模型
         count = len(info)
         for i in range(1, count-1):
             if i == 1:
                 self.model.add(Dense(info[i][0], activation=info[i][1], input_dim=input_dim, kernel_initializer=info[i][2]))  # 输入层，Dense表示BP层
             else:
                 self.model.add(Dense(info[i][0], activation=info[i][1], kernel_initializer=info[i][2]))
 
         # 添加输出层
         self.model.add(Dense(info[count-1][0], activation=info[count-1][1], kernel_initializer=info[count-1][2]))
 
         sgd = SGD(lr=info[0][0], decay=1e-6, momentum=0.9, nesterov=True)
         self.model.compile(loss='binary_crossentropy',  optimizer=sgd,  metrics=['accuracy'])  # 编译模型
 
         self.model.fit(x_train, y_train, validation_data=(x_test, y_test), epochs=info[0][1], batch_size=info[0][2])  # 训练模型1000次
 
         scores_train = self.model.evaluate(x_train, y_train, batch_size=10)
         scores_test = self.model.evaluate(x_test, y_test, batch_size=10)
         scores = self.model.evaluate(data_x, data_y, batch_size=10)
 
         self.finish_trigger.emit()         # 循环完毕后发出信号
         list = [scores_train[1]*100, scores_test[1]*100, scores[1]*100]
         self.evaluate_trigger.emit(list)
         result = pd.Series(self.model.predict(data_x).T[0])
         result.name = '预测(BP)'
         self.result_trigger.emit(result)
         K.clear_session()  # 反复调用model 模型
 
     def save_model(self, save_dir):
         self.model.save(save_dir)  # 保存模型

GUI显示代码（部分）：

 class MainWindow(QtGui.QMainWindow):
     save_dir_signal = QtCore.pyqtSignal(str)  # 传递保存目录信号
 
 def show_evaluate_result(self, evaluate_result):
         help = QtGui.QMessageBox.information(self, '评价结果',
                                              "训练集正确率:  %.2f%%\n测试集正确率:  %.2f%%\n数据集正确率:  %.2f%%" %
                                              (evaluate_result[0], evaluate_result[1], evaluate_result[2]),
                                              QtGui.QMessageBox.Yes)
 
         self.pop_save_dir()
 
     def pop_save_dir(self):
         msg = QtGui.QMessageBox.information(self, '提示', '是否保存模型？', QtGui.QMessageBox.Yes | QtGui.QMessageBox.No)
         if msg == QtGui.QMessageBox.Yes:
                 save_dir = QtGui.QFileDialog.getSaveFileName(self, '选择保存目录', 'C:\\Users\\fuqia\\Desktop')
 
                 if save_dir != '':
                     save_dir = save_dir + '.model'
                     self.save_dir_signal.emit(save_dir)
 
     def show_bp_result(self, result):
 
         self.predict_data = result
         TableWidgetDeal.add_predict_data(self.table, result)
 
     def waiting_label_close(self):
         self.label.close()
 
     def show_waiting(self):
         self.label = QtGui.QLabel(self)
         self.label.setFixedSize(640, 480)  # 不加的话有问题？？？
         self.label.setWindowFlags(QtCore.Qt.FramelessWindowHint)  # 无边框
         self.label.setAttribute(QtCore.Qt.WA_TranslucentBackground)  # 背景透明
 
         screen = QtGui.QDesktopWidget().screenGeometry()
         size = self.label.geometry()
         # 如果是self.label.move((screen.width() - size.width()) / 2 , (screen.height() - size.height()) / 2)无法居中
         self.label.move((screen.width() - size.width()) / 2 + 240, (screen.height() - size.height()) / 2)
 
         # 打开gif文件
         movie = QtGui.QMovie("./Icon/waiting.gif")
         # 设置cacheMode为CacheAll时表示gif无限循环，注意此时loopCount()返回-1
         movie.setCacheMode(QtGui.QMovie.CacheAll)
         # 播放速度
         movie.setSpeed(100)
         self.label.setMovie(movie)
         # 开始播放，对应的是movie.start()
         movie.start()
         self.label.show()
         q = QtCore.QEventLoop()
         q.exec_()

 w = WorkThread()
 w.init(self.object.data_set, feature, label, self.bp_ui.bp_info)
 w.start()
 w.finish_trigger.connect(self.waiting_label_close)
 w.result_trigger.connect(self.show_bp_result)
 w.evaluate_trigger.connect(self.show_evaluate_result)
 self.save_dir_signal.connect(w.save_model)
 self.show_waiting()