Stochastic Deep Supervision for Gallery-Consistent Early-Exit Face Recognition

Researchers propose SD-ViT, a training framework that gives Vision Transformers a usable early exit at every block depth for face recognition, addressing the fixed-depth inference that limits deployment of ViTs in resource-constrained settings. The framework makes every depth produce discriminative embeddings, according to the paper's abstract. The work targets gallery-consistent early-exit face recognition, though the source excerpt provides no benchmark figures or named authors.

Vision Transformers ViTs perform strongly in face recognition FR , yet their fixed-depth inference limits deployment in resource-constrained settings. We propose SD-ViT, a training framework that gives a usable exit at every block depth by making every depth produce discriminative embeddings, for