# Partial FC: Training 10 Million Identities on a Single Machine

arXiv: https://arxiv.org/abs/2010.05222 · HTML: https://mlsanity.com/paper/2010.05222/

## Key facts

- Cited by 163 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 11 Oct 2020.

## Built on by (most cited first, 30 of 163)

- [Towards Metrical Reconstruction of Human Faces](https://mlsanity.com/paper/2204.06607/) (2022)
- [ElasticFace: Elastic Margin Loss for Deep Face Recognition](https://mlsanity.com/paper/2109.09416/) (2021)
- [A Survey on Long-Tailed Visual Recognition](https://mlsanity.com/paper/2205.13775/) (2022)
- [Learning to Relight Portrait Images via a Virtual Light Stage and Synthetic-to-Real Adaptation](https://mlsanity.com/paper/2209.10510/) (2022)
- [BlendFace: Re-designing Identity Encoders for Face-Swapping](https://mlsanity.com/paper/2307.10854/) (2023)
- [YOLO5Face: Why Reinventing a Face Detector](https://mlsanity.com/paper/2105.12931/) (2021)
- [SphereFace2: Binary Classification is All You Need for Deep Face Recognition](https://mlsanity.com/paper/2108.01513/) (2021)
- [WebFace260M: A Benchmark for Million-Scale Deep Face Recognition](https://mlsanity.com/paper/2204.10149/) (2022)
- [Masked Face Recognition Challenge: The InsightFace Track Report](https://mlsanity.com/paper/2108.08191/) (2021)
- [SwinFace: A Multi-task Transformer for Face Recognition, Expression Recognition, Age Estimation and Attribute Estimation](https://mlsanity.com/paper/2308.11509/) (2023)
- [KeyPoint Relative Position Encoding for Face Recognition](https://mlsanity.com/paper/2403.14852/) (2024)
- [MixFaceNets: Extremely Efficient Face Recognition Networks](https://mlsanity.com/paper/2107.13046/) (2021)
- [Unicom: Universal and Compact Representation Learning for Image Retrieval](https://mlsanity.com/paper/2304.05884/) (2023)
- [VideoGen-of-Thought: Step-by-step generating multi-shot video with minimal manual intervention](https://mlsanity.com/paper/2503.15138/) (2025)
- [FaceXFormer: A Unified Transformer for Facial Analysis](https://mlsanity.com/paper/2403.12960/) (2024)
- [PersonalVideo: High ID-Fidelity Video Customization without Dynamic and Semantic Degradation](https://mlsanity.com/paper/2411.17048/) (2024)
- [Unsupervised Face Recognition using Unlabeled Synthetic Data](https://mlsanity.com/paper/2211.07371/) (2022)
- [Benchmarking Algorithmic Bias in Face Recognition: An Experimental Approach Using Synthetic Faces and Human Evaluation](https://mlsanity.com/paper/2308.05441/) (2023)
- [A Multi-modal and Multi-task Learning Method for Action Unit and Expression Recognition](https://mlsanity.com/paper/2107.04187/) (2021)
- [MFR 2021: Masked Face Recognition Competition](https://mlsanity.com/paper/2106.15288/) (2021)
- [TransFace++: Rethinking the Face Recognition Paradigm with a Focus on Accuracy, Efficiency, and Security](https://mlsanity.com/paper/2308.10133/) (2023)
- [Adversarial Mask: Real-World Universal Adversarial Attack on Face Recognition Model](https://mlsanity.com/paper/2111.10759/) (2021)
- [Part-based Face Recognition with Vision Transformers](https://mlsanity.com/paper/2212.00057/) (2022)
- [FaceQAN: Face Image Quality Assessment Through Adversarial Noise Exploration](https://mlsanity.com/paper/2212.02127/) (2022)
- [MagicID: Hybrid Preference Optimization for ID-Consistent and Dynamic-Preserved Video Customization](https://mlsanity.com/paper/2503.12689/) (2025)
- [Multi-modal Emotion Estimation for in-the-wild Videos](https://mlsanity.com/paper/2203.13032/) (2022)
- [Vec2Face: Scaling Face Dataset Generation with Loosely Constrained Vectors](https://mlsanity.com/paper/2409.02979/) (2024)
- [50 Years of Automated Face Recognition](https://mlsanity.com/paper/2505.24247/) (2025)
- [Dynamic Class Queue for Large Scale Face Recognition In the Wild](https://mlsanity.com/paper/2105.11113/) (2021)
- [Leveraging Diffusion For Strong and High Quality Face Morphing Attacks](https://mlsanity.com/paper/2301.04218/) (2023)

## Abstract (arXiv)

Training face recognition models with millions of identities is challenging because classifier storage, logit memory, and computation grow linearly with the number of classes, eventually making full softmax impractical even when the backbone itself fits comfortably in memory. We present Partial FC (PFC), a scalable approximation to large-class softmax that preserves every positive class center while activating only a sampled subset of negative centers in each mini-batch. This asymmetric treatment retains every target term while avoiding exhaustive interaction with millions of mostly uninformative negatives. Our distributed implementation partitions the classifier across GPUs and samples within each owned shard, so sampling reduces local matrix multiplication and logit storage while sharding eliminates class-gradient synchronization across workers. Together, these properties reduce GPU-resident classifier memory, computation, and class-dependent communication without feature-based hard-negative retrieval. End-to-end system benchmarks demonstrate efficient scaling to massive class spaces, including 64 million classes on a single eight-GPU machine. Across large-scale face-recognition datasets, moderate sampling rates maintain competitive recognition accuracy while substantially improving training efficiency. Our best PFC configurations achieve 97.2% TAR on IJB-C at FAR =10-4 and 94.0% TAR on ICCV21-MFR at FAR =10-6. Beyond clean training data, PFC is robust to inter-class conflicts, label noise, and long-tailed identity distributions: under 40% label noise, PFC with conflict filtering raises ICCV21-MFR TAR from 43.9% to 80.2%, while on long-tailed data PFC improves TAR from 87.4% to 92.0%. These results establish positive-preserving negative sampling as an effective foundation for scalable, accurate, and robust identity classification.