Applying Face Recognition on Smart Family Album Player

Abstract

The past decade has seen large advances in technology allowing for technologies only available to governments and large scale companies to now be accessed by anyone with a phone. One such example is the access to facial recognition software with apps like Snapchat, Instagram, and TikTok identifying faces and applying filters based on the user’s request. These apps make it as simple as swiping a screen and immediately the user’s face is recognized and an additional layer is added on top. Facial recognition has capabilities beyond applying filters as it allows people to quickly sort and group images based on a person. Moreover, facial recognition has the flexibility of being added to any previous programming project from beforehand. Therefore, to give users an easy method of sorting images, we intend to create a simple graphical-user interface (GUI) that can sort through hundreds of images without human input. In this poster, we present the design and function of a GUI that would work on many different devices with little set-up. At this moment, the GUI is programmed with $\text{C}++$ and the facial recognition engine is programmed with Python. The graphical-user interface was already programmed in $\text{C}++$ on a Windows device from a previous project before being used for this one. Python was the chosen programming language for facial recognition software as there are large sources of troubleshooting information and it can be implemented in a C++ program without much difficulty. The program as a whole would work through the user specifying a folder of images they have and the program would go through each file in the folder, including files inside other folders, and cluster them based on the people identified. For instance, the folder chosen contains photos of a family of a dad, mom, and a child. From there, each photo is clustered together under the respective person. The libraries used in both the GUI and facial recognition program will be shown and explained on the poster. At the current stage, the facial recognition software is mostly complete as it is able to identify and recognize faces if given training data. In short, the program works by analyzing the commonalities of the training images and determining if the other images match with the data. Currently, the software is being worked on to allow for unsupervised machine learning, meaning the user doesn’t need to give training data from beforehand. Furthermore, integrating the current Python program into the GUI is being looked at. With this project, we hope that any user will find it simple and efficient to sort their memorable family photos without spending any of their own precious time.