ACADSTAFF UGM
| Title | : | Age Estimation on Human Face Image Using Support Vector Regression and Texture-Based Features |
| Author | : |
Jesy S Amelia (1) Wahyono, Ph.D. (2) |
| Date | : | 31 2022 |
| Keyword | : | age detection,LBPBSIF, LBQ, MAE, PCA, preprocessing, feature extraction, support vector regression (SVR) age detection,LBPBSIF, LBQ, MAE, PCA, preprocessing, feature extraction, support vector regression (SVR) |
| Abstract | : | Age estimation is one of the most challenging and crucial issues in the utilization of the facial area to produce useful information. Age estimation can be used in solving scientific problems as well as in subproblems of facial recognition. Age detection can be used by shopping centers to determine the advertising strategies based on the age of visitors and provide recommendations.There are several steps for age detection: preprocessing, feature extraction and predicted of age. In this research, there are three feature extraction methods used and the combination. These include LBP, LPQ and BSIF as well as BSIF+LBP, BSIF+LPQ, and LPQ+LBP and for age detection use SVR Method. MAE is used to calculate the predicted error value. In addition, PCA is used in reducing features resulting from the extraction process. For the dataset, the face-age.zip and UTK Face datasets are 15,202 facial image data which are divided into training 12,162 image and testing 3,040 image. PCA used in the form of 40, 50, 60, 70, 80, and 90 features and the following results are obtained: the combination of extraction methods between BSIF+LBP, BSIF+LPQ and LPQ+LBP gives better results with lower MAE levels than only using one method. The lowest MAE when combining feature extraction using BSIF+LPQ with PCA 70 with first strategy in age estimation MAE 9.766 and second strategy with MAE 9.754. |
| Group of Knowledge | : | Ilmu Komputer |
| Original Language | : | English |
| Level | : | Internasional |
| Status | : |
Published
|
| No | Title | Action |
|---|