The field of face recognition and reconstruction is rapidly evolving, with a focus on improving the accuracy, robustness, and privacy of these systems. Recent developments have led to the creation of more efficient and effective algorithms for face recognition, including the use of diffusion models and cross-spectral matching. Additionally, there is a growing emphasis on protecting facial privacy, with methods such as adversarial identity manipulation and model inversion attacks being explored. Noteworthy papers include: Diffusion-Driven Universal Model Inversion Attack for Face Recognition, which proposes a training-free diffusion-driven universal model inversion attack for face recognition systems. xEdgeFace: Efficient Cross-Spectral Face Recognition for Edge Devices, which presents a lightweight yet effective heterogeneous face recognition framework. Pixel3DMM: Versatile Screen-Space Priors for Single-Image 3D Face Reconstruction, which proposes a set of highly-generalized vision transformers for 3D face reconstruction from a single RGB image.
Advancements in Face Recognition and Reconstruction
Sources
Diffusion-Driven Universal Model Inversion Attack for Face Recognition
Exploiting Multiple Representations: 3D Face Biometrics Fusion with Application to Surveillance
xEdgeFace: Efficient Cross-Spectral Face Recognition for Edge Devices
Pixels2Points: Fusing 2D and 3D Features for Facial Skin Segmentation
A Test Suite for Efficient Robustness Evaluation of Face Recognition Systems
Diffusion-based Adversarial Identity Manipulation for Facial Privacy Protection
A simple and effective approach for body part recognition on CT scans based on projection estimation
The Invisible Threat: Evaluating the Vulnerability of Cross-Spectral Face Recognition to Presentation Attacks
Pixel3DMM: Versatile Screen-Space Priors for Single-Image 3D Face Reconstruction