A total of eight distinct classes are present in the training set representing eight different age groups. Due to the unavailability of GPUs and the extra RAM required for big-data training, I took only 900 approx images for each class to train the model (by distinctly augmenting them, wherever was needed). So, about 7000 photos have been taken for training an age detection model in this project.
METHODOLOGY:- After performing several pre-processing tasks on the data, detected the facial regions from the images and trained a VGG network on them for age recognition!!
The folders with the overall training and testing data can be found here - Train Dataset - https://bit.ly/3pUDDO1 , Test Dataset - https://bit.ly/3iHvwmr