Year: 2025
Type: article
Abstract: Now a days we are facing environmental risks on face landmark detection and it is a more important task thus it can be rectified using simple hardware such as computer vision for widespread applications such as face recognition, expression analysis, and human computer interaction. Visualization to demonstrate face landmark detection on huge environmental conditions remains a goal challenges. This paper presents a lightweight computer vision pipeline that integrates Media Pipe?s, Face Mesh with OpenCV preprocessing techniques to enhance robustness against pose variation, lighting inconsistency, and occlusion. The proposed method classify 3 stages pipeline on histogram equalization, gamma correction, Haar cascade classifier for face detection, and edge-based occlusion to improve landmark detection as real world unconstrained environments. Experimental result shows correct effective accuracy of face landmark as face mesh 90% face landmark detection over real-world to avoid face in security against environmental risk and make world more secure using over face Technology.
Author: K Gayathri
Institution: Anna University, Chennai
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: diamond
OpenAlex ID: W4417459119