VIGFace: Virtual Identity Generation Model for Face Image Synthesis
The author proposes VIGFace, a framework for generating synthetic facial images to address challenges faced by traditional datasets in face recognition. By incorporating virtual prototypes into the model, VIGFace ensures unique virtual identities and achieves state-of-the-art results.
LAFS: Landmark-based Facial Self-supervised Learning for Face Recognition
The author presents LAFS, a landmark-based self-supervised learning approach, to enhance face recognition performance by leveraging facial landmarks and self-supervised pretraining.
Innovative RGB-D Face Recognition Method with Virtual Depth Synthesis
The author presents a novel approach to RGB-D face recognition by leveraging virtual depth synthesis and adaptive confidence weighting for improved accuracy and robustness.
Explainable Face Verification via Feature-Guided Gradient Backpropagation Study
The author introduces the Feature-Guided Gradient Backpropagation method to enhance explainability in face verification systems, providing precise saliency maps for "Accept" and "Reject" decisions.
Contrastive Learning for Facial Action Unit Detection
The author proposes a contrastive learning method to address the scarcity of AU annotations by learning from unlabelled facial videos, achieving discriminative AU representations and person-independent detection.
Boosting Facial Action Unit Detection Through Jointly Learning Facial Landmark Detection and Domain Separation and Reconstruction
The author proposes a new AU detection framework that combines multi-task learning, facial landmark detection, and domain separation to enhance performance in detecting facial action units in the wild.
Efficient Face Recognition with Small Data and Low-Bit Precision
This paper introduces an efficiency-driven approach to face recognition model quantization, demonstrating outstanding results with a smaller dataset. The core argument is that effective quantization can be achieved with minimal data and training time.
