Face Recognition System

In this project, we develop AI model to detect and recognize the face for ID verification purpose. The system is composed with face and its landmark detection model and face recognition model.

We exploit Object Detection and Pose Estimation model to detect face and its landamrk including eye, nose, and mouth for the purpose of higher accuracy. Among many models, we decide to employ pose estimation model from YOLOv11 since it runs with face detection and pose estimation simultaneously.

To increase and ease of face recognition, we applied image alignment for face image from face detection model.

We developed and tested many different feature extraction model for backbone and classification layer for the head of the architecture. Moreover, we employed different loss function. Finally, we build the model with LResnet for backbone, linear layer for head, and subcenter arcface for loss function.