"For computers, distinguishing one face from another is tough. Fluctuating factors like glasses, beards, makeup, hairstyles, and low lighting can easily trip up the system, whereas people can see right through all of that. Until now, humans have outperformed face recognition systems when comparing two images and determining if they are the same person. However, according to the Physics arXiv Blog, a new algorithm called GaussianFace can beat humans. While people can distinguish faces with 97.53% accuracy, the algorithm has reached 98.52%. The system could make accessing smartphones or logging on to computer games with face verification possible."
No comments:
Post a Comment