|
|
| |
| Age and Gender Prediction and Validation Through Single User Images Using CNN |
Abdullah M. Abu Nada, Eman Alajrami, Ahemd A. Al-Saqqa, Samy S. Abu-Naser
|
Abstract:
Recently, the user's gender and age range are very important for organizations to understand their customers 'needs and develop their strategies to provide more enhanced services to them. These organizations mostly rely on their enterprise systems to collect data from users, which forms play an important role in it. This data should be correct and accurate and sometimes must be within a specific format. So, the application form validators and rules were proposed. When we have a look at the related works form validators, we found that their methods work well with the different types of data. But unfortunately, there is a lack of auto detection of the user's age and gender. In addition, they neither detect nor validate the real user's age range that reflects from his/her photo. For example, the ID photo must reflect the person's age range. In this paper, we suggest a new approach to validate the user's gender and age range that is reflected from his photo correctly. Also, adding a double-check layer validator by linking between user photo, gender, and date of birth form inputs based on the Deep Learning approach, by detecting the gender and estimate the age from a single person's photo using a Convolutional Neural Network (CNN or ConvNets). Then, on top of that, a web service to make the validation process is implemented. Finally, we evaluated this solution using the University of Palestine students' photos, and show it has achieved a good result in gender prediction and has challenges in age prediction.
|
|
|
| | |