# RefineFace: Refinement Neural Network for High Performance Face Detection

arXiv: https://arxiv.org/abs/1909.04376 · HTML: https://mlsanity.com/paper/1909.04376/

## Key facts

- Cited by 21 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 10 Sep 2019.

## Built on by (most cited first, 21 of 21)

- [YOLO5Face: Why Reinventing a Face Detector](https://mlsanity.com/paper/2105.12931/) (2021)
- [Disentangle Your Dense Object Detector](https://mlsanity.com/paper/2107.02963/) (2021)
- [YOLO-FaceV2: A Scale and Occlusion Aware Face Detector](https://mlsanity.com/paper/2208.02019/) (2022)
- [Going Deeper Into Face Detection: A Survey](https://mlsanity.com/paper/2103.14983/) (2021)
- [ASFD: Automatic and Scalable Face Detector](https://mlsanity.com/paper/2201.10781/) (2022)
- [CRFace: Confidence Ranker for Model-Agnostic Face Detection Refinement](https://mlsanity.com/paper/2103.07017/) (2021)
- [EResFD: Rediscovery of the Effectiveness of Standard Convolution for Lightweight Face Detection](https://mlsanity.com/paper/2204.01209/) (2022)
- [FrameHopper: Selective Processing of Video Frames in Detection-driven Real-Time Video Analytics](https://mlsanity.com/paper/2203.11493/) (2022)
- [EfficientFace: An Efficient Deep Network with Feature Enhancement for Accurate Face Detection](https://mlsanity.com/paper/2302.11816/) (2023)
- [FDLite: A Single Stage Lightweight Face Detector Network](https://mlsanity.com/paper/2406.19107/) (2024)
- [Pleno-Generation: A Scalable Generative Face Video Compression Framework with Bandwidth Intelligence](https://mlsanity.com/paper/2502.17085/) (2025)
- [Generative Models at the Frontier of Compression: A Survey on Generative Face Video Coding](https://mlsanity.com/paper/2506.07369/) (2025)
- [Balanced Masked and Standard Face Recognition](https://mlsanity.com/paper/2110.01521/) (2021)
- [EfficientSRFace: An Efficient Network with Super-Resolution Enhancement for Accurate Face Detection](https://mlsanity.com/paper/2306.02277/) (2023)
- [A Comprehensive Survey of Masked Faces: Recognition, Detection, and Unmasking](https://mlsanity.com/paper/2405.05900/) (2024)
- [Structure-Guided Diffusion Models for High-Fidelity Portrait Shadow Removal](https://mlsanity.com/paper/2507.04692/) (2025)
- [Wide Aspect Ratio Matching for Robust Face Detection](https://mlsanity.com/paper/2103.05993/) (2021)
- [Bounding-box deep calibration for high performance face detection](https://mlsanity.com/paper/2110.03892/) (2021)
- [Being Aware of Localization Accuracy By Generating Predicted-IoU-Guided Quality Scores](https://mlsanity.com/paper/2309.13269/) (2023)
- [Fully Quantized Always-on Face Detector Considering Mobile Image Sensors](https://mlsanity.com/paper/2311.01001/) (2023)
- [Secure and Efficient UAV-Based Face Detection via Homomorphic Encryption and Edge Computing](https://mlsanity.com/paper/2507.09860/) (2025)

## Abstract (arXiv)

Face detection has achieved significant progress in recent years. However, high performance face detection still remains a very challenging problem, especially when there exists many tiny faces. In this paper, we present a single-shot refinement face detector namely RefineFace to achieve high performance. Specifically, it consists of five modules: Selective Two-step Regression (STR), Selective Two-step Classification (STC), Scale-aware Margin Loss (SML), Feature Supervision Module (FSM) and Receptive Field Enhancement (RFE). To enhance the regression ability for high location accuracy, STR coarsely adjusts locations and sizes of anchors from high level detection layers to provide better initialization for subsequent regressor. To improve the classification ability for high recall efficiency, STC first filters out most simple negatives from low level detection layers to reduce search space for subsequent classifier, then SML is applied to better distinguish faces from background at various scales and FSM is introduced to let the backbone learn more discriminative features for classification. Besides, RFE is presented to provide more diverse receptive field to better capture faces in some extreme poses. Extensive experiments conducted on WIDER FACE, AFW, PASCAL Face, FDDB, MAFA demonstrate that our method achieves state-of-the-art results and runs at 37.3 FPS with ResNet-18 for VGA-resolution images.