Upload
marcia-charles
View
218
Download
1
Embed Size (px)
Citation preview
Multi-Eigenspace Learning for Video-Base Face RecognitionMulti-Eigenspace Learning for Video-Base Face RecognitionLiang Liu1, Yunhong Wang2, Tieniu Tan1
1National Laboratory of Pattern Recognition, Institute of Automation, CAS, Beijing, China2School of Computer Science and Engineering, Beihang University, Beijing, China
模式识别国家重点实验室
中国科学院自动化研究所,北京, 100080
National Laboratory of Pattern Recognition
Institute of Automation, CAS, Beijing, P. R. China
ICB 2007The 2nd International Conference on Biometrics
In this paper, we propose a novel online learning method called Multi-Eigenspace Learning which can learn appearance models incrementally from a given video stream. For each subject, we try to learn a few eigenspace models using IPCA (Incremental Principal Component Analysis). Then, these learnt eigenspace models are used for video-based face recognition. Experimental results show that the proposed method can achieve high recognition rate.
Multi-Eigenspace LearningMulti-Eigenspace Learning Eigenspace model description For the training face video stream of each subject, we aim to construct up to K eigenspaces to approximately represent the appearance manifold of that subject. There are four parameters for each eigenspace, namely
where is the center of the eigenspace. U is a matrix whose columns are orthonormal bases of the eigenspace, namely eigenvectors. is a diagonal matrix. Elements along the diagonal are variances for each principal axis, namely eigenvalues. They are arranged in descending order. N is the number of samples to construct the eigenspace.
The algorithm of Multi-Eigenspace Learning.
Experimental ResultsExperimental Results
ConclusionsConclusionsMulti-Eigenspace Learning is proposed to learn face appearance manifold
online without a pre-trained model.
Experimental results show that the proposed method gives better performance than that given by some other online learning methods.
x
Estabilsh an eigenspace
Compute the distance to each
eigenspace
End
1k
IPCAEstablish a new
eigenspace
Remove an eigenspace
No
Yes
Yes
No
?i n 1i i ?k K
2i
1k k
Yes
No
min ?d T
Data set description
There are 36 video sequences from 36 persons respectively.
The number of frames ranges from 236 to 1270.
Use the first half of each sequence for online learning.
Use the second half of each sequence for recognition.
Fig. 2. Typical samples of the videos.
Fig. 1. The flowchart of Multi-Eigenspace Learning.
Method Probabilistic Manifold
Multi-Eigenspace Learning
Recognition rates 92.4% 96.2%
Online learning 34.3s 7.1s
Pre-training 77.3s -
K
Method
6 7 8 9
EMS + transition 90.0 91.9 96.4 91.4
Multi-Eigenspace Learning
97.5 96.7 98.1 96.4
Experiment 1 We compare the proposed algorithm with our previous work called EMS + transition, which is also an online learning method. This method try to learn K eigenspace models for all subjects, with consideration of transition matrix.
AbstractAbstract
Table 1. Average recognition rates (%) of different methods when choosing K = 6, 7 , 8, 9 respectively.
Experiment 2 We also implemented the Probabilistic Manifold online learning algorithm for comparison. In this method, an appearance model is incrementally learnt online using a pre-trained generic model and successive frames from the video.
Table 2. Comparison of the Probabilistic Manifold algorithm and our proposed algorithm. For Multi-Eigenspace Learning, we choose K = 8.