I Face Detection And Recognition System author: Mukund Agarwal supervisor: Professor Nishan Canagarajah Project Thesis submitted in support of the Degree of Bachelor of Engineering in Electronic and Communications Engineering

38 
1 3 . 4 I m p r o v e F a c e R e c o g n i t i o n R a t e
The face recognition success rate can definitely be further improved. The face recognition 
module was not tested to its full extent in section 9.5. The number of Eigen signatures 
(Eigenvectors) can be increased to see if there is any improvement in the performance. 
Alternatively, it can be also checked that if by reducing them is there any significant drop in 
the performance. If there is no change in the performance as such, than by reducing the 
number of signatures some valuable processing time can be saved. 
There are other various available methods which are more accurate. They can be 
implemented but the time taken has to be taken into consideration. The most interesting and 
promising methods are getting a better average face by using twenty shots with different 
lighting and ages
18
& 3D face recognition
19
. 
1 3 . 5 M u l t i p l e S h o t s
Current implementation only uses one face to compare with the database. In this modification 
multiple shots can be taken in sequence of the same face and checked the average closest 
match.
This will mean that object tracking will also be required to make sure that the multiple shots 
are of the same face. Also a higher frame processing rate is required for it to be effective. 
1 3 . 6 I n t e r a c t i v e T r a i n i n g M o d u l e
An interactive training module can be introduced in the system. This will enable the user to tell 
the system that if the result of the face recognition module is correct or not. If it’s not, then 
the user can be provided the option of simply just clicking on a button which will start a 
learning algorithm in the face recognition module. This will also reduce the false positive and 
false negative results.
This approach is currently implemented in iPhoto 09.
This module may be extended to face detection as well which may reduce the false face 
detection rate. 

---

Mukund Agarwal 
Face Detection & Recognition System