Downloads: 2 | Views: 69 | Weekly Hits: ⮙2 | Monthly Hits: ⮙2

Research Paper | Computer Science | India | Volume 12 Issue 11, November 2023

Creating Age, Gender and Ethnicity Predictions through the use of CNN's

Arjun Khullar

Abstract: Humans are remarkably capable at detecting age, gender and ethnicities solely from a picture of an individual's face. For a computer to be able to identify individuals from images of faces would be challenging for many reasons - the variation in proportions and resolution of images, alongside the presence of facial expressions amongst them. Additionally, this type of facial recognition has several applications, including security, entertainment, and virtual reality. In particular, we will be designing an algorithm to predict the age, gender and ethnicity of an individual based solely on a headshot image of them. We were drawn to this particular application by its seeming abstractness and challenge. We used a convolutional neural network for this, as it is able to learn adaptively using facial features, and has been used in the past for similarly purposed algorithms. The algorithm was deployed on a data set of 23,000 images, with 70% being used for training, 10% for validation, and 20% for testing. We achieved an 80% accuracy in predicting age, gender and ethnicity with this algorithm, and have demonstrated its ability to be deployed in a real-world setting.

Keywords: Age Prediction, Gender prediction, Ethnicity prediction, Machine learning, Artificial intelligence

Edition: Volume 12 Issue 11, November 2023,

Pages: 1256 - 1260