Comparative Evaluation of Face Detection Algorithms

Computer vision is shaping a new era with its constant development of SOTA algorithms. One heavily contested sub-field of computer vision is face detection, due to its versatile usage in many fields such as security, medical diagnosis, entertainment, and military applications. As the technology develops, it aims to run faster and more accurately on mobile devices and remote computers. In this paper, we aim to compare a number of the best face recognition algorithms and analyze the performance of each of them by deploying each algorithm on a Jetson Nano Developer Kit. Among the 6 algorithms discussed, Haar_FA1 showed the best performance in terms of precision. On the other hand, OpenCV_dnn had the best results in terms of recall. Lastly, in terms of execution time-as it is an important metric-face detectors that belong to the Haar classifier family dominated the comparison. © 2020 IEEE.