# An Efficient Training Approach for Very Large Scale Face Recognition

arXiv: https://arxiv.org/abs/2105.10375 · HTML: https://mlsanity.com/paper/2105.10375/

## Key facts

- Cited by 12 papers in the map (data through 14 Oct 2026).
- Builds on 32 of its 55 references that are in the map.
- First posted on arXiv 21 May 2021.

## Builds on (most cited first, 30 of 32)

- [Deep Residual Learning for Image Recognition](https://mlsanity.com/paper/1512.03385/) (2015)
- [Very Deep Convolutional Networks for Large-Scale Image Recognition](https://mlsanity.com/paper/1409.1556/) (2014)
- [Distilling the Knowledge in a Neural Network](https://mlsanity.com/paper/1503.02531/) (2015)
- [Momentum Contrast for Unsupervised Visual Representation Learning](https://mlsanity.com/paper/1911.05722/) (2019)
- [Going Deeper with Convolutions](https://mlsanity.com/paper/1409.4842/) (2014)
- [FaceNet: A Unified Embedding for Face Recognition and Clustering](https://mlsanity.com/paper/1503.03832/) (2015)
- [VGGFace2: A dataset for recognising faces across pose and age](https://mlsanity.com/paper/1710.08092/) (2017)
- [SphereFace: Deep Hypersphere Embedding for Face Recognition](https://mlsanity.com/paper/1704.08063/) (2017)
- [MS-Celeb-1M: A Dataset and Benchmark for Large-Scale Face Recognition](https://mlsanity.com/paper/1607.08221/) (2016)
- [Learning Face Representation from Scratch](https://mlsanity.com/paper/1411.7923/) (2014)
- [Residual Attention Network for Image Classification](https://mlsanity.com/paper/1704.06904/) (2017)
- [Multi-Similarity Loss with General Pair Weighting for Deep Metric Learning](https://mlsanity.com/paper/1904.06627/) (2019)
- [WIDER FACE: A Face Detection Benchmark](https://mlsanity.com/paper/1511.06523/) (2015)
- [Additive Margin Softmax for Face Verification](https://mlsanity.com/paper/1801.05599/) (2018)
- [Cross-Age LFW: A Database for Studying Cross-Age Face Recognition in Unconstrained Environments](https://mlsanity.com/paper/1708.08197/) (2017)
- [WebFace260M: A Benchmark Unveiling the Power of Million-Scale Deep Face Recognition](https://mlsanity.com/paper/2103.04098/) (2021)
- [The MegaFace Benchmark: 1 Million Faces for Recognition at Scale](https://mlsanity.com/paper/1512.00596/) (2015)
- [MobileFaceNets: Efficient CNNs for Accurate Real-Time Face Verification on Mobile Devices](https://mlsanity.com/paper/1804.07573/) (2018)
- [Partial FC: Training 10 Million Identities on a Single Machine](https://mlsanity.com/paper/2010.05222/) (2020)
- [Suppressing Uncertainties for Large-Scale Facial Expression Recognition](https://mlsanity.com/paper/2002.10392/) (2020)
- [Region Attention Networks for Pose and Occlusion Robust Facial Expression Recognition](https://mlsanity.com/paper/1905.04075/) (2019)
- [The Devil of Face Recognition is in the Noise](https://mlsanity.com/paper/1807.11649/) (2018)
- [Deeply learned face representations are sparse, selective, and robust](https://mlsanity.com/paper/1412.1265/) (2014)
- [UMDFaces: An Annotated Face Dataset for Training Deep Networks](https://mlsanity.com/paper/1611.01484/) (2016)
- [Loss Function Search for Face Recognition](https://mlsanity.com/paper/2007.06542/) (2020)
- [Semi-Siamese Training for Shallow Face Learning](https://mlsanity.com/paper/2007.08398/) (2020)
- [Suppressing Mislabeled Data via Grouping and Self-Attention](https://mlsanity.com/paper/2010.15603/) (2020)
- [Accelerated Training for Massive Classification via Dynamic Class Selection](https://mlsanity.com/paper/1801.01687/) (2018)
- [Pairwise Relational Networks for Face Recognition](https://mlsanity.com/paper/1808.04976/) (2018)
- [Support Vector Guided Softmax Loss for Face Recognition](https://mlsanity.com/paper/1812.11317/) (2018)

## Built on by (most cited first, 12 of 12)

- [FaceMAE: Privacy-Preserving Face Recognition via Masked Autoencoders](https://mlsanity.com/paper/2205.11090/) (2022)
- [Face Pyramid Vision Transformer](https://mlsanity.com/paper/2210.11974/) (2022)
- [ARoFace: Alignment Robustness to Improve Low-Quality Face Recognition](https://mlsanity.com/paper/2407.14972/) (2024)
- [Deep Learning Based Face Recognition Method using Siamese Network](https://mlsanity.com/paper/2312.14001/) (2023)
- [Autoencoder Based Face Verification System](https://mlsanity.com/paper/2312.14301/) (2023)
- [GIF: Generative Inspiration for Face Recognition at Scale](https://mlsanity.com/paper/2505.03012/) (2025)
- [QCFace: Image Quality Control for boosting Face Representation & Recognition](https://mlsanity.com/paper/2510.15289/) (2025)
- [A Survey of Face Recognition](https://mlsanity.com/paper/2212.13038/) (2022)
- [Unsupervised Deep Learning Image Verification Method](https://mlsanity.com/paper/2312.14395/) (2023)
- [CAFE: Learning to Condense Dataset by Aligning Features](https://mlsanity.com/paper/2203.01531/) (2022)
- [AttFC: Attention Fully-Connected Layer for Large-Scale Face Recognition with One GPU](https://mlsanity.com/paper/2503.06839/) (2025)
- [Sparsity-Inducing Divergence Losses for Biometric Verification](https://mlsanity.com/paper/2606.31664/) (2026)

## Abstract (arXiv)

Face recognition has achieved significant progress in deep learning era due to the ultra-large-scale and welllabeled datasets. However, training on the outsize datasets is time-consuming and takes up a lot of hardware resource. Therefore, designing an efficient training approach is indispensable. The heavy computational and memory costs mainly result from the million-level dimensionality of thefully connected (FC) layer. To this end, we propose a novel training approach, termed Faster Face Classification (F2C), to alleviate time and cost without sacrificing the performance. This method adopts Dynamic Class Pool (DCP) for storing and updating the identities features dynamically, which could be regarded as a substitute for the FC layer. DCP is efficiently time-saving and cost-saving, as its smaller size with the independence from the whole face identities together. We further validate the proposed F2C method across several face benchmarks and private datasets, and display comparable results, meanwhile the speed is faster than state-of-the-art FC-based methods in terms of recognition accuracy and hardware costs. Moreover, our method is further improved by a well-designed dual data loader including indentity-based and instancebased loaders, which makes it more efficient for the updating DCP parameters.