SD-ViT 如何以隨機深度監督實現與人臉庫一致的提前退出人臉識別
Stochastic Deep Supervision for Gallery-Consistent Early-Exit Face Recognition
SD-ViT 透過隨機深度監督,讓 Vision Transformer 的每個區塊深度都能產生具辨識力的人臉嵌入,支援提前退出識別而無須重新計算人臉庫。基準測試中,SD-ViT 與相同骨幹網絡的全深度常規訓練結果相差不超過 0.4 IJB-C 分;以 50% 計算成本運行時,提前退出探測可達自身全深度表現的 96% 以上。從四分之一深度起,沿用全深度校準的單一驗證閾值仍然有效。
Published on Nov 2
Authors:
,
,
,
Abstract
SD-ViT trains Vision Transformers so every block depth yields discriminative face embeddings, enabling early-exit recognition with minimal accuracy loss and no gallery recomputation.
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 budget-selected fixed-depth exits. Standard ViTs supervise intermediate layers only through backpropagation, so their intermediate representations are not directly optimized to form usable identity embeddings. SD-ViT adds a stochastic auxiliary signal: at each iteration a randomly selected intermediate block is passed through the same embedding head as the final layer and trained with a weighted margin-penalty softmax loss. The objective is thus stochastically sampled deep supervision through a shared classifier. Because both loss terms share that head and classifier, intermediate tokens are trained to map into the same embedding space as final-layer tokens. This matters for FR, where matching is by similarity: a probe processed at an early exit can be compared directly against full-depth gallery embeddings, with no gallery recomputation or adaptation layers. On established benchmarks, SD-ViT stays within 0.4 IJB-C points of the same backbone trained conventionally at full depth, while its early-exit probes recover over 96% of their own full-depth performance at 50% of the computational cost without modifying the reference gallery. A single verification threshold calibrated at full depth remains usable from quarter depth onward.
View arXiv page View PDF GitHub Add to collection
Get this paper in your agent:
hf papers read 2611.09987
Don't have the latest CLI?
curl -LsSf https://hf.co/cli/install.sh | bash
Models citing this paper 0
No model linking this paper
Cite arxiv.org/abs/2611.09987 in a model README.md to link it from this page.
Datasets citing this paper 0
No dataset linking this paper
Cite arxiv.org/abs/2611.09987 in a dataset README.md to link it from this page.
Spaces citing this paper 0
No Space linking this paper
Cite arxiv.org/abs/2611.09987 in a Space README.md to link it from this page.
Collections including this paper 1
來源:HuggingFace Daily Papers(社區熱門論文) · huggingface.co