Automatic facial validation system using real-time semantic segmentation with BiSeNet V2
Multidisciplinary Science Journal · 06.07.2026 · 9 (1) · 2027027
DOI: 10.31893/multiscience.2027027 ↗Kurz gefasst
Fachartikel in der Zeitschrift Multidisciplinary Science Journal, online erschienen am 06.07.2026. Verfasst von Jose Carlos Navarro-Vega und 4 weiteren Autor:innen. Die Arbeit nennt 46 Literaturangaben.
Abstract (Originalsprache)
Student photograph validation is a critical process in issuing student ID cards and academic and administrative documents in higher education institutions. However, in some institutions it is still carried out manually, making it labor intensive, time-consuming, and prone to human error. The main objective of this paper an automated facial validation system that complies with the technical and morphological standards, based on real-time semantic segmentation using the Bilateral Segmentation Network (BiSeNet) V2 model. The solution also includes an interoperability module and a prototype digital student ID card to reduce validation time and support integration with institutional platforms. The methodology was organized into stages: collecting and normalizing student data, automated image processing, role and user management, parameter checking using PyTorch models, and synchronization with institutional systems via a REST API. The system is designed around a three-tier architecture (presentation, application, and data) using PostgreSQL, FastAPI, and ReactJS to ensure scalability and modularity. The BiSeNet V2 model, implemented in PyTorch, performs real-time facial segmentation, identifies facial features, and verifies the required technical and morphological parameters. The results showed a 96,9% reduction in validation time compared with the manual process, cutting processing time from 60 hours to under 2 for a set of 416 photographs. In addition, the system was capable of automatically correcting and validating photographs presenting deviations in technical parameters (e.g., resolution, size, or format), achieving a validation accuracy exceeding 80% in such cases. Interoperability via REST API facilitated integration with institutional systems, such as the academic system. A prototype digital student ID card was also developed, featuring a QR code linked to the student’s identity. Overall, this proposal provides a modern, scalable, and efficient solution for university identity management, reducing human error and improving institutional efficiency.
Zitieren
APA
Navarro-Vega, J. C., Gutierrez, E. P., Cordova, N. A. M., Tambo, S. J. S., & Castillo, R. M. (2026). Automatic facial validation system using real-time semantic segmentation with BiSeNet V2. Multidisciplinary Science Journal, 9(1), 2027027. https://doi.org/10.31893/multiscience.2027027
Vancouver
Navarro-Vega JC, Gutierrez EP, Cordova NAM, Tambo SJS, Castillo RM. Automatic facial validation system using real-time semantic segmentation with BiSeNet V2. Multidisciplinary Science Journal. 2026;9(1):2027027. doi:10.31893/multiscience.2027027
BibTeX
@article{NavarroVega2026,
author = {Navarro-Vega, Jose Carlos and Gutierrez, Elias Paredes and Cordova, Nick Angel Macedo and Tambo, Samuel Jose Sanchez and Castillo, Ralph Miranda},
title = {Automatic facial validation system using real-time semantic segmentation with BiSeNet V2},
journal = {Multidisciplinary Science Journal},
year = {2026},
volume = {9},
number = {1},
pages = {2027027},
doi = {10.31893/multiscience.2027027}
}
EUVIMED