Creat membership Creat membership
Sign in

Forgot password?

Confirm
  • Forgot password?
    Sign Up
  • Confirm
    Sign In
Creat membership Creat membership
Sign in

Forgot password?

Confirm
  • Forgot password?
    Sign Up
  • Confirm
    Sign In
Collection

toTop

If you have any feedback, Please follow the official account to submit feedback.

Turn on your phone and scan

home > search >

Improving the age estimation accuracy by a hybrid optimization scheme

Author:
Ghufran, Reza Syahroel  Leu, Jenq-Shiou  Prakosa, Setya Widyawan  


Journal:
MULTIMEDIA TOOLS AND APPLICATIONS


Issue Date:
2018


Abstract(summary):

Age estimation from digital contents is an interesting topic. Face image reading for age estimation is an intuitive way after classifying face images into several predefined age groups. Age estimation can be regarded as a multiclass problem due to the variation of given individuals determined by genes and several external factors. There are various applications from age estimation, such as forensic and image detection. To approximate and detect the age of each person, the implementation of age estimation algorithm is essential. Many proposed algorithms contain several methods to build the whole mechanism. However, the implementation of SVM combined with optimization is not explored yet. Using an optimization algorithm supposedly can improve accuracy of age estimation algorithm. Based on that fact, we apply GA and PSO for this goal due to their simplicity and powerful ability. GA and PSO are used for this study to make the comparison between both optimizations. This work derives the results for both with optimization and without optimization to prove whether the approach has been successfully realized or not. In the proposed scheme, we conducted age range estimation with five predefined classes and combined several techniques of extracting estimation information from image data, such as Local Gabor Binary Patterns (LGBP) for filtering, Principle Component Analysis (PCA), Linear Discriminant Analysis (LDA) for feature extraction, Support Vector Machines (SVM) for classification, as well as Genetic Algorithm (GA) and Particle Swarm Optimization (PSO) for optimization. Genetic Algorithm and Particle Swarm Optimization were used to find the most suitable parameters to carry out the SVM method. Experimental results show that our proposed hybrid method can raise the estimation accuracy compared to other schemes. The outcome of fold 3 is enhanced up to 14% for both GA and PSO, and the scheme with GA has diminished processing time up to 26 s while it with PSO can reduce up to 25 s.


Page:
2543---2559


Similar Literature

Submit Feedback

This function is a member function, members do not limit the number of downloads