机器学习代码学习(六)-＞＞人脸表情辨识别

- 日理万妓 2022-04-22 12:46 143阅读 0赞

人脸表情辨识作为本科毕设的项目，部分代码如下：

# 人脸表情辨识 #

下面代码为: .ipynb文件, 可以在Jupyter上运行  
题目为:  
给了2个.csv文件(一个训练,一个预测)  
求:人脸表情是哪种

import numpy as np#处理矩阵
    import pandas as pd#资料处理(读csv文件)
    import tensorflow as tf
    import matplotlib.pyplot as plt#画图
    
    #读入文件
    %matplotlib inline
    train_data= pd.read_csv('/Users/liyixin/Desktop/AI/fourth/train.csv')#训练
    test_data= pd.read_csv('/Users/liyixin/Desktop/AI/fourth/test.csv')#测试
    
    # print(train_data['feature'])#里面为数据
    # print(test_data['feature'])
    
    #对train_data进行处理:  将[’feature‘]每行分隔开 ,并存为矩阵形式
    train_data['feature']=train_data['feature'].apply(lambda im: np.fromstring(im,sep=' '))
    x_train=np.vstack(train_data['feature'].values)#所有的train_data图片
    print(x_train)
    #用于显示图片
    x_train1=x_train.reshape(-1,48,48)
    # y_train=train_data['label']
    # print(y_train)
    # print(x_train.shape)
    import matplotlib.pyplot as plt 
    def plot_image(image):  #显示一张图片
        fig = plt.gcf()  
        fig.set_size_inches(2,2)  
        plt.imshow(image, cmap='binary') # cmap='binary' 參數設定以黑白灰階顯示. 
    #     plt.imshow(image)
        plt.show() 
      #显示第0张图片
    plot_image(x_train1[0])
    from keras.utils import to_categorical
    # 对train_data进行处理: 将['label']存成array 并进行onehot编码
    y_train_label=np.array(train_data['label'])
    print(y_train_label)
    # print(y_train.shape)
    
    #one-hot编码
    y_train_onehot= to_categorical(y_train_label)
    print(y_train_onehot)
    #对train_data进行处理后的shape:
    x_train.shape,y_train_onehot.shape
    #对train_data进行处理,前5000作为测试,后面用于训练
    x_valid_image=x_train[0:5000,:]#测试
    x_train_image=x_train[5000:,:]#训练
    y_valid_onehot=y_train_onehot[0:5000,:]
    y_train_onehot=y_train_onehot[5000:,:]
    
    #用于显示图片
    x_valid1_image=x_valid_image.reshape(-1,48,48)#测试图片的显示
    
    
    print(x_valid_image.shape,y_valid_onehot.shape)
    print(x_train_image.shape,y_train_onehot.shape)
    
    label_dict={0:"angry",1:"disgust",2:"fear",3:"happy",4:"sad",5:"suprise",6:"neutral" }
    
    import matplotlib.pyplot as plt 
    def plot_images_labels_predict(images, labels, prediction, idx, num=10):  
        fig = plt.gcf()  
        fig.set_size_inches(12, 14)  
        for i in range(0, num):  
            ax=plt.subplot(5,5, 1+i)  
            ax.imshow(images[idx], cmap='binary')  
            title = str(i)+'——'+label_dict[labels[idx]]
            if len(prediction) > 0:  
                title+='-->'+label_dict[prediction[idx]] 
            ax.set_title(title, fontsize=10)  
            ax.set_xticks([]); ax.set_yticks([])  
            idx+=1  
        plt.show()  
    
    
    plot_images_labels_predict(x_train1,y_train_label,[],0,num=10)
    
    
    from keras.utils import to_categorical
    #对test进行处理: 将[’feature‘] 并存为矩阵形式
    test_data['feature']=test_data['feature'].apply(lambda im: np.fromstring(im,sep=' '))
    x_test=np.vstack(test_data['feature'].values)
    print(x_test)
    print(x_test.shape)
    #用于显示图片
    x_test1=x_test.reshape(-1,48,48)
    
    # 对test_data进行处理: 将['id']存成array 
    y_test_id=np.array(test_data['id'])
    print(y_test_id)
    
    # feature reshape
    # 用于显示图片
    # x_valid1=x_valid.reshape(-1,48,48)
    # print(x_valid1.shape)
    # x_train=x_train.reshape(-1,48,48,1)
    # x_test1=x_test.reshape(-1,48,48,1)
    
    #train_data中的数据和test_data中的数据  reshape 格式转换(float) 标准化
    x_valid_image_norm=x_valid_image.reshape(-1,48,48,1).astype('float32') / 255
    x_train_image_norm=x_train_image.reshape(-1,48,48,1).astype('float32') / 255
    x_test_image_norm=x_test.reshape(-1,48,48,1).astype('float32') / 255
    
    print(x_valid_image_norm.shape,x_train_image_norm.shape,x_test_image_norm.shape)
    print(x_valid_image_norm[0])
    # 
    from keras.models import Sequential  
    from keras.layers import Dense,Dropout,Flatten,Conv2D,MaxPooling2D  
      
    model = Sequential()  
    # Create CN layer 1  
    model.add(Conv2D(filters=16,kernel_size=(3,3),padding='same',input_shape=(48, 48, 1),activation='relu'))  
    # Create Max-Pool 1  
    model.add(MaxPooling2D(pool_size=(2,2)))  
    # Create CN layer 2  
    model.add(Conv2D(filters=36,kernel_size=(3,3),padding='same',activation='relu'))  
    # Create Max-Pool 2  
    model.add(MaxPooling2D(pool_size=(2,2)))  
    # Create CN layer 3 
    model.add(Conv2D(filters=64,kernel_size=(3,3),padding='same',activation='relu'))  
    # Create Max-Pool 3 
    model.add(MaxPooling2D(pool_size=(2,2)))  
    # # Add Dropout layer 
    model.add(Dropout(0.5)) 
    model.add(Flatten())  
    model.add(Dense(128, activation='relu'))  
    model.add(Dropout(0.5)) 
    model.add(Dense(7, activation='softmax'))  
    
    model.summary()  
    print("")  
    from keras import optimizers
    
    # 定義訓練方式  
    model.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['accuracy'])  
      
    # 開始訓練  
    train_history = model.fit(x=x_train_image_norm,y=y_train_onehot, validation_split=0.2,epochs=200, batch_size=300, verbose=1) 
    import matplotlib.pyplot as plt  
    def show_train_history(train_history, train, validation):  
        plt.plot(train_history.history[train])  
        plt.plot(train_history.history[validation])  
        plt.title('Train History')  
        plt.ylabel(train)  
        plt.xlabel('Epoch')  
        plt.legend(['train', 'validation'], loc='upper left')  
        plt.show()
    
    
    show_train_history(train_history,'acc','val_acc')
    
    show_train_history(train_history,'loss','val_loss')
    
    scores = model.evaluate(x=x_valid_image_norm,#model.evaluate检测(100笔数据)->用于测试准确度(0.81)
                            y=y_valid_onehot)
    print(scores[1])
    prediction = model.predict_classes(x_valid_image_norm)  # Making prediction and save result to prediction
    # print(prediction)
    print("\t[Info] Show 10 prediction result (From valid image):")  
    print("%s\n" % (prediction[0:10]))
    #显示valid图片的 原label和 预测后的label
    plot_images_labels_predict(x_valid1_image,y_train_label[0:5000],prediction,0,num=10) 
    print("\t[Info] Display Confusion Matrix:")#混淆矩陣 (Confusion matrix) 
    import pandas as pd  
    print("%s\n" % pd.crosstab(y_train_label[0:5000], prediction, rownames=['label'], colnames=['predict']))  
    import matplotlib.pyplot as plt 
    def show_predict_probability(x_image,y,prediction,predicted_probability, i):  
        print(i,'valid_image','label:',label_dict[y[i]],'predict',label_dict[prediction[i]])
        plt.figure(figsize=(2,2))
        plt.imshow(x_image[i]) 
        plt.show()  
        for j in range(7):  
             print(label_dict[j]+' probability:%1.9f'%(predicted_probability[i][j]))
             prediction_probability = model.predict(x_valid_image_norm)#第0张图片
    show_predict_probability(x_valid1_image,y_train_label,prediction,prediction_probability,0)
    import matplotlib.pyplot as plt 
    def show_images_predict(images,prediction, idx, num=10):  
        fig = plt.gcf()  
        fig.set_size_inches(12, 14)  
        for i in range(0, num): 
    #         plt.figure(figsize=(6,6))
            plt.imshow(images[i]) 
            ax=plt.subplot(5,5,1+i)  
            ax.imshow(images[idx], cmap='binary')  
            title='predict'+'-->'+label_dict[prediction[idx]] 
            ax.set_title(title, fontsize=10)  
            ax.set_xticks([]); ax.set_yticks([])  
            idx+=1  
        plt.show()  
        predict_result=model.predict_classes(x_test_image_norm) 
        print("%s\n" % (predict_result[0:10]))
    #显示test图片前十张 的 预测后的label
    show_images_predict(x_test1,predict_result[0:10],0,num=10) s=({"id":test_data["id"],"Label":predict_result})
    submit=pd.DataFrame(data=s)
    submit.to_csv('/Users/liyixin/Desktop/AI/fourth/result.csv',index=False)