Age and Gender Prediction Using Convolutional Neural Network

Abstract

Automatic age and gender classification has been widely used in a large amount of applications, particularly in human–computer interaction, biometrics, visual surveillance, electronic customer and commercial applications. Predicting age and gender of humans is very difficult and complicated. Since the increase in use of social media, autonomous prediction of age and gender has become extremely important. Nevertheless, performance of the existing methods on real-world images is still significantly lacking, especially when compared to the tremendous leaps in performance recently reported for the related task of face recognition. In this research paper, we have observed an increase in the performance on these tasks by using convolutional neural networks (CNN). We collected a dataset from ‘Google’ and ‘IMFDB.’ This dataset consisted of about 100,000 images of male and female of different age-groups (5–60). We created a CNN network adjoining the fully connected network using ‘Keras’ library. Following that, three convoluted layers were inserted; the first layer consisting of 64 neurons and window of 7 × 7, followed by a ReLU activation layer and a max-pooling layer of 2 × 2. Two more CNN layers of 100 and 64 neurons were inserted, followed by windows of 5 × 5 and 3 × 3, respectively, with activation function as ReLU. An output was received in matrix form which was flattened using a flatten layer. The layer was then fed to fully connected layers with 64 and 1 neurons each. The final output layer consisted of a single neuron with sigmoid activation function.