Age prediction with partial facial covering

Abstract

Covid-19 is still raging around the world, and the behaviour of humans wearing masks in public will continue to exist. The detection or recognition of faces wearing masks has gradually become a popular research direction at this stage. At the same time, age detection is also a forward-looking research topic. As more applications or websites collect customers' age information with the permission of customers, intending to push content belonging to their age groups to customers of different ages. This dissertation explores the feasibility of age prediction for faces wearing masks. The research is based on the Python language environment. The research starts with face detection, extracts the area containing face information in a large environment, and displays the results in the designed GUI interface through the trained age prediction model. Throughout the experiment, I used two improved convolutional neural network models to train three age classifications, and selected the one that best met the criteria and placed it in the display program.