Year: 2025
Type: article
Abstract: Applications of 3D face reconstruction include biometric authentication, personalized avatars and digital identity, medical visualization, forensic analysis, and broader human-computer interaction. We propose an approach to 3D face reconstruction that can generate a fully textured 3D facial model using only two grayscale images: a front view and a profile view of the subject. Once trained, the system can perform the reconstruction autonomously without manual intervention. Unlike traditional methods requiring multi-camera setups, depth sensors, or cloud-based processing, the proposed approach runs fully offline on a standard CPU, supporting dynamic execution across CPU cores and eliminating the need for a dedicated GPU. A Conditional Deep Generative Adversarial Network (CDGAN) merges identity features from both input images, and the final 3D output is exported in OBJ/MTL format. The method reduces average model size by approximately 70%, enabling faster rendering and easier storage on devices with limited resources. This demonstrates that high-quality 3D face reconstruction can be achieved from minimal input and autonomous CPU-based neural processing, providing a foundation for secure, accessible, and privacy-preserving on-device 3D vision technology.
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7134518334
FWCI:
Cites:
Cited by:
Related to: