pca-face-recognition
所属分类:图形图像处理
开发工具:MultiPlatform
文件大小:4066KB
下载次数:3
上传日期:2019-04-29 20:20:50
上 传 者:
不知可信
说明: 一个简单的人脸识别演示使用PCA算法。pca_svm_face_recogition.m是一种Matlab代码,它利用PCA实现人脸识别程序,降低特征的维数,并利用one-vs-one多类SVM对图像进行分类。
我使用PCA将数据缩减到50维,然后使用SVM线性核函数进行分类,最终得到了0.9437的准确率。
(A simple face recognition demonstration using PCA algorithm. Pca_svm_face_recogition.m is a kind of MATLAB code. It uses PCA to realize face recognition program, reduces the dimension of features, and uses one-vs-one multi-class SVM to classify images.
I use PCA to reduce the data to 50 dimensions, then use SVM linear kernel function to classify, and finally get the accuracy of 0.9437.)
文件列表:
pca_face_recognition.py (2283, 2018-07-09)
__MACOSX (0, 2019-04-29)
._pca_face_recognition.py (212, 2018-07-09)
pca_svm_face_recogition.m (2598, 2018-07-09)
._pca_svm_face_recogition.m (212, 2018-07-09)
orl_faces (0, 2018-07-09)
orl_faces\s34 (0, 2018-07-09)
orl_faces\s34\7.pgm (10318, 2018-07-09)
orl_faces (0, 2019-04-29)
orl_faces\s34 (0, 2019-04-29)
orl_faces\s34\._7.pgm (212, 2018-07-09)
orl_faces\s34\6.pgm (10318, 2018-07-09)
orl_faces\s34\._6.pgm (212, 2018-07-09)
orl_faces\s34\4.pgm (10318, 2018-07-09)
orl_faces\s34\._4.pgm (212, 2018-07-09)
orl_faces\s34\5.pgm (10318, 2018-07-09)
orl_faces\s34\._5.pgm (212, 2018-07-09)
orl_faces\s34\1.pgm (10318, 2018-07-09)
orl_faces\s34\._1.pgm (212, 2018-07-09)
orl_faces\s34\2.pgm (10318, 2018-07-09)
orl_faces\s34\._2.pgm (212, 2018-07-09)
orl_faces\s34\3.pgm (10318, 2018-07-09)
orl_faces\s34\._3.pgm (212, 2018-07-09)
orl_faces\s34\8.pgm (10318, 2018-07-09)
orl_faces\s34\._8.pgm (212, 2018-07-09)
orl_faces\s34\9.pgm (10318, 2018-07-09)
orl_faces\s34\._9.pgm (212, 2018-07-09)
orl_faces\s34\10.pgm (10318, 2018-07-09)
orl_faces\s34\._10.pgm (212, 2018-07-09)
orl_faces\._s34 (212, 2018-07-09)
orl_faces\s33 (0, 2018-07-09)
orl_faces\s33\7.pgm (10318, 2018-07-09)
orl_faces\s33 (0, 2019-04-29)
orl_faces\s33\._7.pgm (212, 2018-07-09)
orl_faces\s33\6.pgm (10318, 2018-07-09)
orl_faces\s33\._6.pgm (212, 2018-07-09)
... ...
# pca-face-recognition
A simple face recognition demo using PCA algorithm.
# Use KNN classifier
Using the ORL face database.
I set ```k = 90```, and the accuracy reach 92.5%.
```shell
$ python pca_face_recognition.py
s1/10.pgm is the most similar to s1/5.pgm
s2/10.pgm is the most similar to s2/8.pgm
s3/10.pgm is the most similar to s3/9.pgm
s4/10.pgm is the most similar to s4/5.pgm
s5/10.pgm is the most similar to s40/5.pgm
s6/10.pgm is the most similar to s6/6.pgm
s7/10.pgm is the most similar to s7/5.pgm
s8/10.pgm is the most similar to s8/3.pgm
s9/10.pgm is the most similar to s9/5.pgm
s10/10.pgm is the most similar to s8/6.pgm
s11/10.pgm is the most similar to s11/1.pgm
s12/10.pgm is the most similar to s12/9.pgm
s13/10.pgm is the most similar to s13/5.pgm
s14/10.pgm is the most similar to s14/1.pgm
s15/10.pgm is the most similar to s15/2.pgm
s16/10.pgm is the most similar to s16/3.pgm
s17/10.pgm is the most similar to s17/7.pgm
s18/10.pgm is the most similar to s18/5.pgm
s19/10.pgm is the most similar to s19/3.pgm
s20/10.pgm is the most similar to s20/2.pgm
s21/10.pgm is the most similar to s21/8.pgm
s22/10.pgm is the most similar to s22/9.pgm
s23/10.pgm is the most similar to s23/1.pgm
s24/10.pgm is the most similar to s24/9.pgm
s25/10.pgm is the most similar to s25/3.pgm
s26/10.pgm is the most similar to s26/8.pgm
s27/10.pgm is the most similar to s27/1.pgm
s28/10.pgm is the most similar to s28/1.pgm
s29/10.pgm is the most similar to s29/3.pgm
s30/10.pgm is the most similar to s30/2.pgm
s31/10.pgm is the most similar to s31/5.pgm
s32/10.pgm is the most similar to s32/9.pgm
s33/10.pgm is the most similar to s33/2.pgm
s34/10.pgm is the most similar to s34/6.pgm
s35/10.pgm is the most similar to s35/5.pgm
s36/10.pgm is the most similar to s36/6.pgm
s37/10.pgm is the most similar to s37/9.pgm
s38/10.pgm is the most similar to s38/5.pgm
s39/10.pgm is the most similar to s39/6.pgm
s40/10.pgm is the most similar to s5/1.pgm
accuracy: 0.925000
```
# Use SVM classifier
```pca_svm_face_recogition.m``` is a Matlab code which implements a face recognition program using PCA to reduce the dimension of the features and one-vs-one multiclass SVM to classify the image.
I used PCA to reduce the data to 50 dimensions and then use SVM linear kernel function to classify, finally, I got an accuracy of 0.9437.
```
accuracy =
0.9437
```
Here is the eigen faces:
![eigen_faces.jpg](https://github.com/techping/pca-face-recognition/raw/master/eigen_faces.jpg)
近期下载者:
相关文件:
收藏者: