GIF: Generative Inspiration for Face Recognition at Scale

TL;DR

This paper introduces GIF, a method that replaces scalar labels with structured identity codes in face recognition, reducing computational complexity from linear to logarithmic while improving accuracy.

Contribution

The paper proposes a novel tokenization scheme and training approach that predicts structured identity codes, significantly lowering computational costs in large-scale face recognition.

Findings

Outperforms competitors by 1.52% and 0.6% at TAR@FAR=1e-4 on IJB-B and IJB-C.

Transforms computational cost growth from linear to logarithmic with respect to identity count.

Demonstrates improved scalability and accuracy in face recognition benchmarks.

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 https://github.com/msed-Ebrahimi/GIF