UCM 分类

import cv2#读取图片需要
import numpy as np
from sklearn.metrics import classification_report
from sklearn.model_selection import train_test_split
from torchvision.datasets import ImageFolder
dataset = ImageFolder('C:\\Users\\LENOVO\\Desktop\\Images\\')
a = cv2.imread('C:/Users/LENOVO/Desktop/Images/0/agricultural00.tif')
dataset = dataset.imgs
print(dataset)
data_path = []
data_label = []
for i in range(dataset.__len__()):
    data_path.append(dataset[i][0])
    data_label.append(dataset[i][1])
#append意味增加元素；[]为列表；dataset.__len__()定义数据集长度
data_arr=[]
for i in range(len(data_path)):
    p = cv2.imread(data_path[i])
    p = cv2.resize(p,(224,224))
#p为输入的原图像；dsize输出图像所需大小
    data_arr.append(np.array(p).flatten())
print(type(data_arr))
data_arr  = np.array(data_arr)#转换成向量
data_label = np.array(data_label)
X_train, X_test, y_train, y_test = train_test_split(data_arr, data_label, test_size=0.1, random_state=200)
from sklearn import svm
predictor = svm.SVC(gamma='scale', C=1.0,max_iter = 1000)
#错误术语的惩罚参数C;核函数系数，float类型，可选参数，默认为auto。
# 进行训练
predictor.fit(X_train, y_train)
predictions_labels = predictor.predict(X_test)
print(classification_report(y_test, predictions_labels,digits=3))
#是一种经典的监督学习算法，用于解决二分类和多分类问题
#其核心思想是通过在特征空间中找到一个最优的超平面来进行分类，并且间隔最大