简单的人脸识别

导包

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
%matplotlib inline
from pandas import DataFrame
from sklearn.decomposition import PCA
from sklearn.svm import SVC
from sklearn.model_selection import train_test_split
from sklearn.model_selection import GridSearchCV
from sklearn.datasets import fetch_lfw_people

faces = fetch_lfw_people(resize=1,min_faces_per_person=70)

# 查看一下数据
images = faces.images
plt.imshow(images[100],cmap='gray')
display(images.shape,images[100].shape)

#获取数据
data = faces.data
target = faces.target
target_name = faces.target_names

# 切分数据
X_train ,X_test,y_train,y_test = train_test_split(data,target,test_size=88)

# 使用PCA降维
pca = PCA(100,whiten=True)
pca.fit(X_train)
# 转化
X_train_pca = pca.transform(X_train)

使用GridSearchCV找到最佳的支持向量机方法

svc = SVC()
param_grid = {
    'C':[0.01,0.1,1,2,10],
    'gamma':[0.001,0.01,0.1,0.5,1]
}
gv = GridSearchCV(svc,param_grid)
gv.fit(X_train_pca,y_train)

# 最佳参数
gv.best_params_
# 最好的得分
gv.best_score_
# 最佳的模型
estimator = gv.best_estimator_

# 对预测数据也进行降维
X_test_pca = pca.transform(X_test)
# 最佳模型进行预测
y_ = estimator.predict(X_test_pca)

预测得分

estimator.score(X_test_pca,y_test)
> 0.8181818181818182

图形可视化

# 画图
#  88 / 40
# 8行5列
plt.figure(figsize=(8*2,5*6))
for i in range(40):
    axes = plt.subplot(8,5,i+1)
    axes.imshow(X_test[i*2].reshape((125,94)),cmap='gray')
    axes.set_title('True:%s\nPredict:%s' % (target_name[y_test[i*2]].split()[-1],target_name[y_[i*2]].split()[-1]))
    axes.axis('off')

1.png 2.png 3.png