Abstract :

With the proliferation of social media and platforms, automatic age and gender categorization has gained relevance in a growing number of applications. When compared to the recent reported great jumps in performance for the related job of face recognition, the performance of present approaches on real-world photographs is still severely insufficient. We demonstrate in this research that deep-convolution neural network (CNN) models (age_net.caffemodel, gender_net.caffe model) can learn representations and significantly improve performance on these tasks. So, even with a little quantity of training data, we suggest a basic convolution net design.