Gender Classification and Age Prediction using CNN and ResNet in Real-Time

The face is the most dominant part of the human body, we can get a lot of information from facial features such as detecting the face of person, gender classification and even age prediction. In current times, Computer Vision (CV) has been used to train machines to comprehend and envision the real world. In this research, a novel artifact has been presented to detect face, classify genders, and predict age from the human facial images all in real-time using a live stream from a camera source. Convolutional Neural Networks (CNN) have been used for training purposes along with the CV library, Keras. To power this novel study, each model has been trained separately and finely tuned before merging them into the final system. The use of the careful modern architecture of CNN and current regularisation methods have been properly evaluated and implemented. The accuracy of the developed model has been calculated manually achieving an overall accuracy of 85%. All the testing has been performed in real-time. After extensive testing and evaluation, a state of the art novel system has been developed with a combination of simple pre-processing steps. This system can be broadly deployed for security purposes, as in airports and police checkpoints, and also to restrict the access of alcohol from vending machines to under-aged people.