基于SVM实现焊接缺陷图片识别分类Python源码(含数据集，需自己训练模型).zip

import numpy as np import os import re import sklearn from sklearn.svm import SVC path_of_TRAINDATA = '/home/wufan/Desktop/缺陷图像集/VectorsForTrain/' path_of_TESTDATA = '/home/wufan/Desktop/缺陷图像集/VectorsForTest/' #四个字段的正则表达式 bar_defect = r'\u6761\u5f62\u7f3a\u9677' unpenetrated = r'\u672a\u710a\u900f' unfused = r'\u672a\u7194\u5408' circular_defect = r'\u5706\u5f62\u7f3a\u9677' vectors = [] #预定存储图像转换的向量 labels = [] #预定存储图像的标签 x_test = [] #测试集向量 y_test = [] #测试集标签 for root, dirs, files in os.walk(path_of_TRAINDATA): for file in files: f = open(path_of_TRAINDATA + file, 'r') #打开txt文件，只读 lines = f.readlines() for line in lines: vec = line.split() #将txt文件里按空格分开的数字存入数组vec a = [] for ele in vec: ele = float(ele) #将数组vec里的string型数字改为float a.append(ele) vectors.append(a) #用正则表达式匹配来输入标签数组 if re.search(re.compile(bar_defect), file): #条形缺陷记作0 labels.append(0) elif re.search(re.compile(unpenetrated), file): #未焊透记作1 labels.append(1) elif re.search(re.compile(unfused), file): #未熔合记作2 labels.append(2) elif re.search(re.compile(circular_defect), file): #圆形缺陷记作3 labels.append(3) else: labels.append(9) filename = [] #将测试数据的文件名记录，与分类结果一起打印输出 #将测试数据存入数组x_test for root, dirs, files in os.walk(path_of_TESTDATA): for file in files: f = open(path_of_TESTDATA + file, 'r') lines = f.readlines() for line in lines: vec = line.split() a = [] for ele in vec: ele = float(ele) a.append(ele) filename.append(file) x_test.append(a) #将实际的标签存入数组y_test for file in filename: if re.search(re.compile(bar_defect), file): #条形缺陷记作0 y_test.append(0) elif re.search(re.compile(unpenetrated), file): #未焊透记作1 y_test.append(1) elif re.search(re.compile(unfused), file): #未熔合记作2 y_test.append(2) elif re.search(re.compile(circular_defect), file): #圆形缺陷记作3 y_test.append(3) else: #随便写的 y_test.append(9) #支持向量机 Vectors = np.array(vectors) #训练集向量，要转换成numpy.array型向量 Labels = np.array(labels) #训练集标签 X_Test = np.array(x_test) #测试集向量 Y_Test = np.array(y_test) #测试集标签 clf = SVC(gamma = 'auto') #分类面 clf.fit(Vectors, Labels) #训练分类面，具体参数参考https://www.cnblogs.com/luyaoblog/p/6775342.html SVC(C=1.0, cache_size=200, class_weight=None, coef0=0.0, decision_function_shape='ovr', degree=3, gamma='auto', kernel='rbf', max_iter=-1, probability=False, random_state=None, shrinking=True, tol=0.001, verbose=False) print(filename) #打印测试集文件名 print(clf.predict(X_Test)) #打印预测结果 print('训练集精度：' + str(100 * clf.score(Vectors, Labels)) + '%') #打印精度 print('测试集精度：' + str(100 * clf.score(X_Test, Y_Test)) + '%')