A LIGHTWEIGHT AND MULTI - BRANCH MODULE IN FACIAL SEMANTIC SEGMENTATION FEATURE EXTRACTION
DOI:
https://doi.org/10.62643/Abstract
Facial semantic segmentation is a fundamental task in computer vision that focuses on accurately identifying and separating different facial regions such as the eyes, nose, mouth, hair, and skin at the pixel level. It plays a vital role in applications including facial recognition, biometric authentication, augmented reality, emotion analysis, surveillance, and medical image processing. However, many existing segmentation models require high computational resources and large numbers of parameters, making them less suitable for real-time and resource-constrained environments. This project proposes A Lightweight and Multi-Branch Module in Facial Semantic Segmentation Feature Extraction, which is based on the U-Net architecture. The proposed model incorporates multi-scale feature extraction, channel attention mechanisms, and depthwise separable convolutions to improve segmentation accuracy while significantly reducing computational complexity. The encoder–decoder architecture with skip connections preserves both high-level semantic information and fine spatial details, enabling accurate facial feature extraction. The system follows a systematic workflow consisting of facial image acquisition, preprocessing, feature extraction, semantic segmentation, and segmentation mask generation. Experimental results demonstrate that the proposed lightweight model achieves reliable pixellevel segmentation with fewer parameters and lower computational cost compared to conventional approaches. The proposed system is suitable for deployment in real-time applications and low-resource devices while maintaining high segmentation performance across different facial images and environmental conditions.
Downloads
Published
Issue
Section
License

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.













