FaceGCD: Generalized Face Discovery via Dynamic Prefix Generation

2025-07-29

AI Summary

Traditional face recognition is confined to closed-set identity classification and struggles to discover unknown identities in open-world scenarios. To address this, we propose Generalized Face Discovery (GFD), a novel task that unifies known-identity recognition—supporting both labeled and unlabeled identities—with unknown-identity discovery for the first time. Existing generalized category discovery (GCD) methods fail on faces due to their high cardinality and fine-grained inter-class similarities. To overcome this, we introduce a dynamic prefix generation mechanism: a lightweight HyperNetwork generates instance-specific, layer-shared prefixes that adaptively modulate the backbone feature extractor, eliminating reliance on large static models. Evaluated on the GFD benchmark, our method significantly outperforms state-of-the-art GCD approaches and the ArcFace baseline, achieving new SOTA performance. This work establishes a principled framework for open-world face recognition.

Technology Category

• Computer Vision: Generative Adversarial Networks (GANs) for Vision
• Machine Learning: Multi-class/Multi-label Learning & Extreme Classification
• Search and Optimization: Mixed Discrete/Continuous Search

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Recognizing and differentiating among both familiar and unfamiliar faces is a critical capability for face recognition systems and a key step toward artificial general intelligence (AGI). Motivated by this ability, this paper introduces generalized face discovery (GFD), a novel open-world face recognition task that unifies traditional face identification with generalized category discovery (GCD). GFD requires recognizing both labeled and unlabeled known identities (IDs) while simultaneously discovering new, previously unseen IDs. Unlike typical GCD settings, GFD poses unique challenges due to the high cardinality and fine-grained nature of face IDs, rendering existing GCD approaches ineffective. To tackle this problem, we propose FaceGCD, a method that dynamically constructs instance-specific feature extractors using lightweight, layer-wise prefixes. These prefixes are generated on the fly by a HyperNetwork, which adaptively outputs a set of prefix generators conditioned on each input image. This dynamic design enables FaceGCD to capture subtle identity-specific cues without relying on high-capacity static models. Extensive experiments demonstrate that FaceGCD significantly outperforms existing GCD methods and a strong face recognition baseline, ArcFace, achieving state-of-the-art results on the GFD task and advancing toward open-world face recognition.

Problem

• Unify face identification with generalized category discovery
• Recognize labeled and unlabeled faces while discovering new identities
• Address challenges of high cardinality and fine-grained face IDs

Innovation

• Dynamic prefix generation for feature extraction
• HyperNetwork adaptively outputs prefix generators
• Lightweight layer-wise prefixes for fine-grained IDs