HOG-PCA-SVM

HOG+SVM+PCA的训练及测试代码，使用的是sklearn框架，人脸数据集为lfw_funneled，最后效果：PCA+SVM:85，HOG+SVM:94,SVM+PCA+HOG:92

各部分输出：（1）PCA+SVM Total dataset size: n_samples: 1288 n_features: 1850 n_classes: 7 Extracting the top 100 eigenfaces from 966 faces done in 0.435s Projecting the input data on the eigenfaces orthonormal basis (966, 100) (322, 100) done in 0.018s Fitting the classifier to the training set done in 23.820s Best estimator found by grid search: SVC(C=1000.0, class_weight='balanced', gamma=0.005) 5 Predicting people's names on the test set done in 0.039s [3] 3 precision recall f1-score support Ariel Sharon 0.67 0.77 0.71 13 Colin Powell 0.83 0.90 0.86 60 Donald Rumsfeld 0.85 0.63 0.72 27 George W Bush 0.89 0.97 0.93 146 Gerhard Schroeder 0.90 0.76 0.83 25 Hugo Chavez 0.90 0.60 0.72 15 Tony Blair 0.87 0.75 0.81 36 accuracy 0.86 322 macro avg 0.84 0.77 0.80 322 weighted avg 0.86 0.86 0.86 322 [[ 10 0 1 2 0 0 0] [ 1 54 0 4 0 1 0] [ 3 2 17 5 0 0 0] [ 0 4 0 142 0 0 0] [ 0 1 1 1 19 0 3] [ 0 3 0 1 1 9 1] [ 1 1 1 5 1 0 27]]

可以看到，训练模型的时间为23.820s，训练结果为平均精度达到了0.86，混淆矩阵显示1号（样本13人）检测到14人，其中正确的人数为10人，另有1人为2号，3人为3号，正确的1号样本有1人被检测成了3号，2人被检测成了4号。（2）HOG+SVM train done in 1739.726s Best estimator found by grid search: SVC(C=1000.0, cache_size=200, class_weight='balanced', coef0=0.0, decision_function_shape='ovr', degree=3, gamma=0.0001, kernel='rbf', max_iter=-1, probability=False, random_state=None, shrinking=True, tol=0.001, verbose=False) precision recall f1-score support

 Ariel Sharon       1.00      0.85      0.92        13
 Colin Powell       0.94      1.00      0.97        60

Donald Rumsfeld 0.92 0.81 0.86 27 George W Bush 0.94 0.99 0.96 146 Gerhard Schroeder 0.92 0.96 0.94 25 Hugo Chavez 1.00 0.67 0.80 15 Tony Blair 0.94 0.83 0.88 36

  avg / total       0.94      0.94      0.94       322

[[ 11 0 1 1 0 0 0] [ 0 60 0 0 0 0 0] [ 0 1 22 2 1 0 1] [ 0 0 1 145 0 0 0] [ 0 0 0 1 24 0 0] [ 0 1 0 3 0 10 1] [ 0 2 0 3 1 0 30]]

可以看到，训练模型的时间为1739.726s（大概半个小时），平均准确度达到0.94。从混淆矩阵中可以看到1号与6号出现了明显的过拟合现象。（3）HOG+PCA+SVM train done in 17.062s Best estimator found by grid search: SVC(C=1000.0, cache_size=200, class_weight='balanced', coef0=0.0, decision_function_shape='ovr', degree=3, gamma=0.005, kernel='rbf', max_iter=-1, probability=False, random_state=None, shrinking=True, tol=0.001, verbose=False)

               precision    recall  f1-score   support

 Ariel Sharon       1.00      0.85      0.92        13
 Colin Powell       0.90      1.00      0.94        60

Donald Rumsfeld 0.79 0.81 0.80 27 George W Bush 0.93 0.97 0.95 146 Gerhard Schroeder 0.96 0.88 0.92 25 Hugo Chavez 0.91 0.67 0.77 15 Tony Blair 0.97 0.81 0.88 36

  avg / total       0.92      0.92      0.91       322

[[ 11 0 1 1 0 0 0] [ 0 60 0 0 0 0 0] [ 0 1 22 2 1 0 1] [ 0 0 5 141 0 0 0] [ 0 1 0 1 22 1 0] [ 0 1 0 4 0 10 0] [ 0 4 0 3 0 0 29]]

可以看到模型的训练时间为17.062s，远小于HOG+SVM的训练速度，从精度方面，平均精度为0.92，从混淆矩阵中可以看到，1号与6号的过拟合现象已经明显减弱。

lvfengkun/HOG-PCA-SVM

HOG-PCA-SVM