Automatic classification of person by age group
Automatic age estimation is attracting increasing research interest recently, mainly due to its wide potential applications, which includes forensic medicine, developing group specific human computer interfaces for human-machine system, improving recognition system efficiency on large databases, per...
Saved in:
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Final Year Project |
| Language: | English |
| Published: |
2010
|
| Subjects: | |
| Online Access: | http://hdl.handle.net/10356/40753 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Nanyang Technological University |
| Language: | English |
| Summary: | Automatic age estimation is attracting increasing research interest recently, mainly due to its wide potential applications, which includes forensic medicine, developing group specific human computer interfaces for human-machine system, improving recognition system efficiency on large databases, performing age specific customer analysis and other commercial purposes. Therefore, it is beneficial to explore the traits related to aging and develop a system to automatically estimate a person’s age.
In earlier researches, face dimension ratios and wrinkles were used to classify the age group of a person. Other works have elected to work directly on the image intensity values of the face as a series of patches or as a spatially normalized holistic image. On average, contemporary methods yield a system performance of 80% for estimation at ±5 years from the actual age.
In this project, the aim is to develop an age estimation system which is able to predict a person’s age within 5 years of deviation from the actual age. The system was implemented by first detecting the prominent facial features like the eyes, nose and mouth, which are often the visual cues for age estimation by human eye using programming functions from OpenCV library. Locality Preserving Projection (LPP) is then used to train the system to learn the age manifold of 713 Asian’s faces between the ages of 6 to 80 years old. This allows the system to extract the manifold of a face image and predict the person’s age. The implemented system yields a good result with an average mean error of 6.1 years.
Detail description of the system is described in this Final Year Project report, together with the results of each process, discussions and recommendations for further improvements. |
|---|