Abstract
Aiming to reduce the computational cost of Softmax in massive label space of Face Recognition (FR) benchmarks, recent studies estimate the output using a subset of identities. Although promising, the association between the computation cost and the number of identities in the dataset remains linear only with a reduced ratio. A shared characteristic among available FR methods is the employment of atomic scalar labels during training. Consequently, the input to label matching is through a dot product between the feature vector of the input and the Softmax centroids. Inspired by generative modeling, we present a simple yet effective method that substitutes scalar labels with structured identity code, i.e., a sequence of integers. Specifically, we propose a tokenization scheme that transforms atomic scalar labels into structured identity codes. Then, we train an FR backbone to predict the code for each input instead of its scalar label. As a result, the associated computational cost becomes logarithmic w.r.t. number of identities. We demonstrate the benefits of the proposed method by conducting experiments. In particular, our method outperforms its competitors by 1.52%, and 0.6% at TAR@FAR$=1e-4$ on IJB-B and IJB-C, respectively, while transforming the association between computational cost and the number of identities from linear to logarithmic. See code at this https URL
Abstract (translated)
为了减少大规模标签空间中人脸识别（FR）基准测试中的Softmax计算成本，最近的研究通过使用身份子集来估计输出。尽管这种方法有前景，但其计算成本与数据集中身份数量之间的关系仍然是线性的，只是比例有所降低。现有可用的FR方法的一个共同特点是，在训练过程中采用原子标量标签。因此，输入和标签匹配是通过输入特征向量与Softmax中心点之间的点积实现的。 受到生成模型的启发，我们提出了一种简单而有效的方法，即用结构化的身份代码替换标量标签，也就是一串整数。具体来说，我们提出了一个标记化方案，将原子标量标签转换为结构化的身份代码。然后，训练FR骨干网络来预测每个输入的身份代码，而不是其标量标签。结果是与身份数量相关的计算成本变为对数关系。 通过实验展示了所提方法的益处：在IJB-B和IJB-C基准测试中，在FAR$=1e-4$下的TAR上，我们的方法分别优于竞争对手1.52%和0.6%，同时将计算成本与身份数量的关系从线性转变为对数。更多详情请参见代码链接：[此处提供实际的URL或指示读者查看原始文档中的具体网址]。 注意：在上述翻译中，“TAR@FAR”是True Acceptance Rate at False Alarm Rate（假警报率下的真实接受率）的缩写，在人脸识别研究领域常用来评估模型性能。
URL
https://arxiv.org/abs/2505.03012
