KNN训练代码

from sklearn.datasets import load_digits
from sklearn.model_selection import train_test_split
import matplotlib.pyplot as plt
from sklearn.neighbors import KNeighborsClassifier
from sklearn.naive_bayes import GaussianNB
from sklearn.linear_model import LogisticRegression
from sklearn.preprocessing import StandardScaler
from sklearn.linear_model import LinearRegression
import cv2
import os
from pandas import DataFrame
###
# 聚类精度模板、三大指标模板
import numpy as np
from scipy.optimize import linear_sum_assignment

def cluster_acc(y_true, y_pred):
    y_true = np.array(y_true).astype(np.int64)
    assert y_pred.size == y_true.size
    D = max(y_pred.max(), y_true.max()) + 1
    w = np.zeros((D, D), dtype=np.int64)
    for i in range(y_pred.size):
        w[y_pred[i], y_true[i]] += 1
    ind = linear_sum_assignment(w.max() - w)
    ind = np.asarray(ind)
    ind = np.transpose(ind)
    return sum([w[i, j] for i, j in ind]) * 1.0 / y_pred.size
#
## 朴素贝叶斯与KNN分类
def knn_gnb_lr_lsr(X, labels, title_knn="XXX.KNN",\
                title_gnb="XXX.GNB",title_lr="XXX.LR",\
                title_lsr="XXX=LSR",n = 3):
    # 划分训练集和测试集
    x_train, x_test, labels_train, labels_test =\
        train_test_split(X, labels, test_size=0.2, random_state=22)

    # 使用KNN进行分类
    knn = KNeighborsClassifier()
    knn.fit(x_train, labels_train)
    label_sample = knn.predict(x_test)
    knn_acc=cluster_acc(labels_test, label_sample)
    print(title_knn,"=",knn_acc)

    # 使用高斯朴素贝叶斯进行分类
    gnb = GaussianNB()  # 使用默认配置初始化朴素贝叶斯
    gnb.fit(x_train, labels_train)  # 训练模型
    label_sample = gnb.predict(x_test)
    gnb_acc = cluster_acc(labels_test, label_sample)
    print(title_gnb,"=", gnb_acc)

    # 线性回归
    lr = LinearRegression()
    lr.fit(x_train, labels_train)
    label_sample = lr.predict(x_test)
    label_sample = np.round(label_sample)
    label_sample=label_sample.astype(np.int64)
    lr_acc = cluster_acc(labels_test, label_sample)
    print(title_lr, "=", lr_acc)

    #Logistic regression 需要事先进行标准化
    #创建一对多的逻辑回归对象
    # 标准化特征
    scaler = StandardScaler()
    X_ = scaler.fit_transform(X,labels)
    # 划分训练集和测试集
    x_train, x_test, labels_train, labels_test = \
        train_test_split(X_, labels, test_size=0.2)
    log_reg = LogisticRegression()#multinomial
    #训练模型
    log_reg.fit(x_train, labels_train)
    label_sample = log_reg.predict(x_test)
    lsr_acc = cluster_acc(labels_test, label_sample)
    print(title_lsr, "=", lsr_acc)

    return round(knn_acc,n),round(gnb_acc,n),round(lr_acc,n),round(lsr_acc,n)

def createDatabase(path,re_size=(36,136)):
    file = os.listdir(path)
    file_number = len(file)
    print("**********************")
    print("file_number",file_number)
    img = []
    labels = []
    for k in range(0, file_number):
        subfile = os.listdir(path + '/' +file[k])
        subfile_number = len(subfile)
        for i in range(0, subfile_number):
            image = cv2.imread(path + '/' + file[k] + '/' + subfile[i], cv2.IMREAD_COLOR)
            image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
            image = cv2.resize(image, re_size)#re_size=(50,50)元组
            image = image.reshape(1,image.size)
            img.append(image)
            labels.append(k)#可更改，以符合数据集
    img = np.array(img)
    labels = np.array(labels)
    return img,labels

def get_img(path,re_size=(70,70)):
    from PIL import Image

    file = os.listdir(path)
    img = []
    labels = []
    for i in range(0, len(file)):
        file_dir = os.path.join(path, file[i])
        image = Image.open(file_dir)
        image = image.resize((re_size), Image.ANTIALIAS)
        # 使用Image模块进行resize好处是图片路径可以包含中文||Image.ANTIALIAS 即是抗锯齿
        img.append(np.array(image))
        labels.append(i//80)  #获取标签，可以根据实际情况更改

    img = np.array(img)
    return img ,labels


digit = []
#face
path = 'E:\\pythonwork\\LPDdetection\\Chinese-license-plate-recognition-system\\Train\\plate4'
X1, Y1 = createDatabase(path,(180,200))
X1 = X1.reshape(200, 180*200*3)
figure = knn1, gnb1,lr1,lsr1 = knn_gnb_lr_lsr(X1, Y1)
digit.append(figure)