This project explores age-invariant face recognition (AIFR), which aims to recognize individuals accurately despite changes in appearance due to aging. AIFR has applications in security, access control, and criminal detection.
- Focused on adapting and evaluating the VGG-Face model for the AIFR task.
- Applied transfer learning techniques by freezing and modifying layers.
- Conducted experiments on the FG-NET and MORPH datasets to analyze performance.
- Demonstrated the effectiveness of VGG-Face in handling age-related variations.
- Provided comparative insights within the larger study that also considered FaceNet and OpenFace.
Published in the Springer Conference Series.
- Title: Deep Features for Age-Invariant Face Recognition
- Link: DOI Link
Shows how transfer learning with VGG-Face can improve robustness in recognizing identities across age progression.
References: https://github.com/mzaradzki/neuralnets