Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
Email:
2
0
2
Achieving a 40-50% reduction in spoof detection errors while maintaining robustness against adversarial attacks, MFAD redefines the standards for explainable face anti-spoofing systems.
Calibrating text embeddings with visual prototypes boosts segmentation accuracy without any additional training, bridging the gap between generic concepts and specific visual instances.