如何在SVM图像分类中保存混淆矩阵的假阳性与假阴性样本

嘿，这事儿好办！我来帮你把那些分类错误的样本单独保存下来，方便你后续分析。先明确两个关键概念：

• 假阳性（FP）：实际是negative类别的图像，却被模型误判成了positive
• 假阴性（FN）：实际是positive类别的图像，却被模型误判成了negative

接下来我们只需要在你的现有代码里加几部分逻辑：判断每个测试样本的真实标签和预测结果，把对应错误的样本复制到指定目录就行。下面是修改后的完整代码，我标注了新增的关键部分：

import sklearn
from sklearn import metrics
from sklearn import svm
from sklearn.svm import SVC
from sklearn.metrics import accuracy_score, f1_score, precision_score, recall_score, classification_report, confusion_matrix
import numpy as np
import PIL
from PIL import Image
import os
import shutil  # 新增：用于复制文件到指定目录

# 新增：定义假阳性和假阴性的保存路径，可根据需求修改
save_false_positive = "C:/false_positive"
save_false_negative = "C:/false_negative"
COMMON_SIZE = (50, 50)

# 新增：确保保存目录存在，不存在则自动创建
for save_dir in [save_false_positive, save_false_negative]:
    if not os.path.exists(save_dir):
        os.makedirs(save_dir)

def image_to_matrix(filename):
    """ takes a filename and turns it into a numpy array of RGB pixels """
    img = Image.open(filename)
    img = img.resize(COMMON_SIZE)
    img = np.asarray(img)
    return img

def flatten_matrix(matrix):
    """ takes in an (m, n) numpy array and flattens it into an array of shape (1, m * n * 3) """
    s = matrix.shape[0] * matrix.shape[1] * 3
    mat = matrix.reshape(1,s)
    return mat[0]

classes = ['positive', 'negative']
# 训练集加载逻辑（原代码不变）
train_image_dirs = ["C:/potholes_50/train/positive/", "C:/potholes_50/train/negative/"]
train_images = []
train_labels = []
for directory in train_image_dirs:
    temp_list = [directory+ f for f in os.listdir(directory)]
    train_labels.extend(len(temp_list)*[classes[train_image_dirs.index(directory)]])
    train_images.extend(temp_list)

train_data = []
for image in train_images:
    image = image_to_matrix(image)
    image = flatten_matrix(image)
    train_data.append(image)
train_data = np.array(train_data)

# 测试集加载逻辑（原代码不变）
test_labels = []
test_image_dirs = ["C:/potholes_50/model_test/positive/", "C:/potholes_50/model_test/negative/"]
test_images = []
for directory in test_image_dirs:
    temp_list = [directory+ f for f in os.listdir(directory)]
    test_labels.extend(len(temp_list)*[classes[test_image_dirs.index(directory)]])
    test_images.extend(temp_list)

test_data = []
for image in test_images:
    image = image_to_matrix(image)
    image = flatten_matrix(image)
    test_data.append(image)
test_data = np.array(test_data)

# SVM训练与预测（原代码不变）
clf = svm.SVC(kernel = 'linear',gamma =5, C=1)
clf.fit(train_data, train_labels)
print("SVM average score: %f" % clf.score(test_data, test_labels))
predicted_label = clf.predict(test_data)

# 评估指标计算（原代码不变）
Accuracy_Score = accuracy_score(test_labels, predicted_label)
Precision_Score = precision_score(test_labels, predicted_label, average="macro")
Recall_Score = recall_score(test_labels, predicted_label, average="macro")
F1_Score = f1_score(test_labels, predicted_label, average="macro")
print('Average Accuracy: %0.2f +/- (%0.1f) %%' % (Accuracy_Score.mean()*100, Accuracy_Score.std()*100))
print('Average Precision: %0.2f +/- (%0.1f) %%' % (Precision_Score.mean()*100, Precision_Score.std()*100))
print('Average Recall: %0.2f +/- (%0.1f) %%' % (Recall_Score.mean()*100, Recall_Score.std()*100))
print('Average F1-Score: %0.2f +/- (%0.1f) %%' % (F1_Score.mean()*100, F1_Score.std()*100))
print("")
CM = confusion_matrix(test_labels, predicted_label)
print(CM)

# 新增：遍历测试样本，保存假阳性和假阴性样本
for img_path, true_label, pred_label in zip(test_images, test_labels, predicted_label):
    # 假阳性：真实标签是negative，预测为positive
    if true_label == 'negative' and pred_label == 'positive':
        shutil.copy(img_path, save_false_positive)
        print(f"已保存假阳性样本: {os.path.basename(img_path)}")
    # 假阴性：真实标签是positive，预测为negative
    elif true_label == 'positive' and pred_label == 'negative':
        shutil.copy(img_path, save_false_negative)
        print(f"已保存假阴性样本: {os.path.basename(img_path)}")

关键修改说明 #

1. 导入shutil模块：用来实现文件复制功能，把错误分类的样本复制到指定目录
2. 创建保存目录：用os.makedirs确保保存假阳性/假阴性的目录存在，避免因目录不存在报错
3. 遍历判断并保存：通过zip同时遍历测试样本路径、真实标签和预测标签，逐一判断样本类型，复制到对应目录

运行这段代码后，你就能在C:/false_positive和C:/false_negative目录下找到所有分类错误的样本了，文件名和原测试集一致，方便你对应查看原始数据~

内容的提问来源于stack exchange，提问作者Vosco