This project focuses on the generation and authentication of fingerprints using Adaptive Deep Convolution Generative Adversarial Networks (ADCGAN). Fingerprints are vital for human identification due to their uniqueness. This project leverages deep learning, specifically ADCGAN, to address the complexities of fingerprint authentication in various applications, from mobile security to airport systems.
- Fingerprint Synthesis: Utilizes ADCGAN to generate realistic fingerprint images.
- Authentication: Implements a model to authenticate generated fingerprints with high accuracy.
- Dataset: Trained on the Socofing dataset, achieving 92% accuracy.
- Applications: Suitable for enhancing security in devices and systems where fingerprint authentication is critical.
Deep learning has revolutionized fields like natural language processing, computer vision, and speech processing. This project explores its application in fingerprint synthesis and biometrics. Generative Adversarial Networks (GANs) are particularly effective in generating realistic samples from existing data distributions. ADCGAN extends this capability, enabling unrestricted fingerprint research by bypassing the confidentiality issues typically associated with biometric data.
For more details, refer to my full paper: Fingerprint Generation and Authentication using ADCGAN
Additionally, you can find more information here: Semantic Scholar