Real Time Facial Recognition and Tracking System Using Drones

Facial recognition technology has come a long way and it is still rapidly advancing. A system composed of a drone capable of autonomous facial detection and recognition, used to find and track a specified individual in real time, is the subject of this paper. The automated system works by having a drone responsible for video capture and streaming, while a computer, receiving that video stream, will run modern day facial recognition algorithms and relay instructions back to the drone. These instructions control the drone's movements with the goal of identifying and tracking a singular specified individual. Three facial recognition systems were implemented and tested to see which works best in this system. The three facial recognition systems being tested are Local Binary Pattern Histogram (LBPH), FaceNet, and Face_Recognition, all of which were chosen due to a mixture of their performance and prominence. Experiments were conducted to quantify the performance of each facial recognition system with respect to the distance of the drone from an individual, while also taking into consideration the angle of an individual's face and common accessories that cause facial obstruction. Additional experiments were conducted to measure each facial recognition system's performance by examining how many frames per second (FPS) each system could analyze. The results showed that the implemented automated drone system that uses facial recognition technology to track an individual in real time is practical and can have huge implications in the security field, such as locating a lost child in a crowd or identifying targets prior to a police raid.