การตรวจจับดวงตาของผู้ปิดหน้ากากสำหรับป้องกันการติดเชื้อจากภาพถ่ายในหลายมุมมอง

Abstract

This engineering project presents the Eye detection under occlusion by facial masks in various directions. The objective of this project is to develop face detection process for general facial mask detections. This research proposed a face detection method by adding an eye detection process next to an input of the whole facial mask detection process. In this research the researcher focused on the eye detection process that based on a Haar-Cascade algorithm, this method can determine the direction of the face, there are 3 types of the result: Front face, Left side and Right side. The result shows that the accuracy of an eye detection from front-face images is 83.33% and the accuracy of an eye detection from side-face images is 6.25% The accuracy of a method from the results is acceptable but still requires more dataset to make sure the accuracy of the proposed method would change in the better way.