Sample and Computation Redistribution for Efficient Face Detection
Not on the map. Why?
Abstract (arXiv)
Although tremendous strides have been made in uncontrolled face detection, efficient face detection with a low computation cost as well as high precision remains an open challenge. In this paper, we point out that training data sampling and computation distribution strategies are the keys to efficient and accurate face detection. Motivated by these observations, we introduce two simple but effective methods (1) Sample Redistribution (SR), which augments training samples for the most needed stages, based on the statistics of benchmark datasets; and (2) Computation Redistribution (CR), which reallocates the computation between the backbone, neck and head of the model, based on a meticulously defined search methodology. Extensive experiments conducted on WIDER FACE demonstrate the state-of-the-art efficiency-accuracy trade-off for the proposed scrfd family across a wide range of compute regimes. In particular, scrfdf34 outperforms the best competitor, TinaFace, by 3.86% (AP at hard set) while being more than 3× faster on GPUs with VGA-resolution images. We also release our code to facilitate future research.
Key facts. Cited by 86 papers in the map (data through 7 Oct 2026). Builds on 34 of its 51 references that are in the map. First posted on arXiv 10 May 2021.
Builds on
34 of 51 references.
References · 34
- 2015Deep Residual Learning for Image Recognitioncited by 51,809
- 2014Microsoft COCO: Common Objects in Contextcited by 17,498
- 2016Feature Pyramid Networks for Object Detectioncited by 5,021
- 2014Going Deeper with Convolutionscited by 4,231
- 2017MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applicationscited by 3,827
- 2015FaceNet: A Unified Embedding for Face Recognition and Clusteringcited by 2,833
- 2019FCOS: Fully Convolutional One-Stage Object Detectioncited by 1,573
- 2018Group Normalizationcited by 1,514
- 2019MMDetection: Open MMLab Detection Toolbox and Benchmarkcited by 1,379
- 2020Designing Network Design Spacescited by 954
- 2018Regularized Evolution for Image Classifier Architecture Searchcited by 902
- 2018Path Aggregation Network for Instance Segmentationcited by 796
22 more
- 2017Focal Loss for Dense Object Detectioncited by 749
- 2019Bridging the Gap Between Anchor-based and Anchor-free Detection via Adaptive Training Sample Selectioncited by 539
- 2017How far are we from solving the 2D & 3D Face Alignment problem? (and a dataset of 230,000 3D facial landmarks)cited by 524
- 2019Distance-IoU Loss: Faster and Better Learning for Bounding Box Regressioncited by 454
- 2018Bag of Tricks for Image Classification with Convolutional Neural Networkscited by 430
- 2019Single Path One-Shot Neural Architecture Search with Uniform Samplingcited by 411
- 2020Generalized Focal Loss: Learning Qualified and Distributed Bounding Boxes for Dense Object Detectioncited by 408
- 2015WIDER FACE: A Face Detection Benchmarkcited by 297
- 2019NAS-FPN: Learning Scalable Feature Pyramid Architecture for Object Detectioncited by 289
- 2018Joint 3D Face Reconstruction and Dense Alignment with Position Map Regression Networkcited by 195
- 2019RetinaFace: Single-stage Dense Face Localisation in the Wildcited by 164
- 2021Revisiting ResNets: Improved Training and Scaling Strategiescited by 158
- 2017S3FD: Single Shot Scale-invariant Face Detectorcited by 142
- 2018DSFD: Dual Shot Face Detectorcited by 115
- 2018PyramidBox: A Context-assisted Single Shot Face Detectorcited by 57
- 2017FaceBoxes: A CPU Real-time Face Detector with High Accuracycited by 44
- 2017SSH: Single Stage Headless Face Detectorcited by 40
- 2020OSTeC: One-Shot Texture Completioncited by 37
- 2020TinaFace: Strong but Simple Baseline for Face Detectioncited by 30
- 2019LFFD: A Light and Fast Face Detector for Edge Devicescited by 21
- 2019Computation Reallocation for Object Detectioncited by 17
- 2019HAMBox: Delving into Online High-quality Anchors Mining for Detecting Outer Facescited by 4
Built on by
Citing papers per year: 2021: 1; 2022: 10; 2023: 11; 2024: 18; 2025: 24; 2026: 22.
86 papers.
Sort:
By area:
- 2023DataComp: In search of the next generation of multimodal datasetscited by 635
- 2024A Survey of Multimodal Large Language Model from A Data-centric Perspectivecited by 73
- 2022Expanding Accurate Person Recognition to New Altitudes and Ranges: The BRIAR Datasetcited by 61
- 2025Insert Anything: Image Insertion via In-Context Editing in DiTcited by 61
- 2023GANDiffFace: Controllable Generation of Synthetic Datasets for Face Recognition with Realistic Variationscited by 49
- 2025Long-Term TalkingFace Generation via Motion-Prior Conditional Diffusion Modelcited by 49
- 2021YOLO5Face: Why Reinventing a Face Detectorcited by 44
- 2025VideoGen-of-Thought: Step-by-step generating multi-shot video with minimal manual interventioncited by 33
- 2025Concat-ID: Towards Universal Identity-Preserving Video Synthesiscited by 28
- 2022There is a Time and Place for Reasoning Beyond the Imagecited by 22
- 2023Towards Realistic Generative 3D Face Modelscited by 20
- 2024InfiniBench: A Benchmark for Large Multi-Modal Models in Long-Form Movies and TV Showscited by 20
38 more
- 2023Boosting Object Detection with Zero-Shot Day-Night Domain Adaptationcited by 17
- 2025ContextGen: Contextual Layout Anchoring for Identity-Consistent Multi-Instance Generationcited by 17
- 2024ExpLLM: Towards Chain of Thought for Facial Expression Recognitioncited by 15
- 2025ComfyGPT: A Self-Optimizing Multi-Agent System for Comprehensive ComfyUI Workflow Generationcited by 14
- 2024Is My Data in Your AI? Membership Inference Test (MINT) applied to Face Biometricscited by 12
- 2024RectifID: Personalizing Rectified Flow with Anchored Classifier Guidancecited by 11
- 20223DMM-RF: Convolutional Radiance Fields for 3D Face Modelingcited by 10
- 2025OmniTalker: One-shot Real-time Text-Driven Talking Audio-Video Generation With Multimodal Style Mimickingcited by 9
- 2025Playmate: Flexible Control of Portrait Animation via 3D-Implicit Space Guided Diffusioncited by 8
- 2024UVMap-ID: A Controllable and Personalized UV Map Generative Modelcited by 7
- 2025Identity-Preserving Image-to-Video Generation via Reward-Guided Optimizationcited by 7
- 2025Exploring Cognitive and Aesthetic Causality for Multimodal Aspect-Based Sentiment Analysiscited by 6
- 2022EResFD: Rediscovery of the Effectiveness of Standard Convolution for Lightweight Face Detectioncited by 5
- 2024Monocular Identity-Conditioned Facial Reflectance Reconstructioncited by 5
- 2022Perspective Reconstruction of Human Faces by Joint Mesh and Landmark Regressioncited by 4
- 2025Virtually Being: Customizing Camera-Controllable Video Diffusion Models with Multi-View Performance Capturescited by 4
- 2022Physically-Based Face Rendering for NIR-VIS Face Recognitioncited by 3
- 2022Whose Emotion Matters? Speaking Activity Localisation without Prior Knowledgecited by 3
- 2023Real-time Multi-person Eyeblink Detection in the Wild for Untrimmed Videocited by 3
- 2023MetaHead: An Engine to Create Realistic Digital Headcited by 3
- 2024Template-based Multi-Domain Face Recognitioncited by 3
- 2025FakeParts: a New Family of AI-Generated DeepFakescited by 3
- 2025Knowledge Completes the Vision: A Multimodal Entity-aware Retrieval-Augmented Generation Framework for News Image Captioningcited by 3
- 2022Learning Audio-Visual embedding for Person Verification in the Wildcited by 2
- 2022Improving Depression estimation from facial videos with face alignment, training optimization and schedulingcited by 2
- 2023Filter-Pruning of Lightweight Face Detectors Using a Geometric Median Criterioncited by 2
- 2023QAFE-Net: Quality Assessment of Facial Expressions with Landmark Heatmapscited by 2
- 2024Double Trouble? Impact and Detection of Duplicates in Face Image Datasetscited by 2
- 2024Inv-Adapter: ID Customization Generation via Image Inversion and Lightweight Adaptercited by 2
- 2024FDLite: A Single Stage Lightweight Face Detector Networkcited by 2
- 2024Towards Unified Facial Action Unit Recognition Framework by Large Language Modelscited by 2
- 2025B-FPGM: Lightweight Face Detection via Bayesian-Optimized Soft FPGM Pruningcited by 2
- 2025Automated Visual Attention Detection using Mobile Eye Tracking in Behavioral Classroom Studiescited by 2
- 2026Towards Spatial Supersensing in the Wildcited by 2
- 2024Boosting Unconstrained Face Recognition with Targeted Style Adversarycited by 1
- 2024CLIP Unreasonable Potential in Single-Shot Face Recognitioncited by 1
- 2024VideoGen-of-Thought: Step-by-step generating multi-shot video with minimal manual interventioncited by 1
- 2025Trajectory-guided Motion Perception for Facial Expression Quality Assessment in Neurological Disorderscited by 1
50 most cited of 86 All 86 (JSON)
- 2026Expression-Diverse References for Identity-Preserving Video Generation8 Oct 2026
- 2026StoryEngine: A State-Grounded Agentic Framework for Video Storytelling27 Sep 2026
- 2026Identity-Centric Video Summarization via Hierarchical Fusion of Biometric, Appearance, and 3D Body Features22 Sep 2026
- 2026Data Diversity, Not Frequency Invariance: A Controlled and Self-Audited Study of Compression-Robust Deepfake Detection26 Aug 2026
- 2026ControlRef: Efficient Layout-Guided Multi-Instance Generation via Anchored 4D-RoPE7 Aug 2026
- 2026Localize, Don't Beautify: Client-Side Control of Image-Editing APIs for Cosmetic Surgery Previews3 Aug 2026
- 2026Diff-ID: Identity Consistent Facial Image Generation and Morphing via Diffusion Models27 Jul 2026
- 2026ID-V2V: Identity-Preserving Video Restylization24 Jul 2026
- 2026Articulated Humanoid Head for a Robot Receptionist Capable of Natural Human Interaction19 Jul 2026
- 2026Towards Spatial Supersensing in the Wild15 Jul 2026
- 2026DriveFace: A Cross-Spectral Through-Glass Face Dataset for On-the-Move Vehicular Border Control15 Jul 2026
- 2026Swapping Faces, Saving Features: A Dual-Purpose Pipeline for Pedestrian Privacy in ITS9 Jul 2026
38 more
- 2026VSRo-200: A Romanian Visual Speech Recognition Dataset for Studying Supervision and Multimodal Robustness9 Jul 2026
- 2026Optimizing Image Preparation and Compression for Face Recognition within 1024 Bytes29 Jun 2026
- 2026TraMP-LLaMA: Generative Interpretability with Decoupled Instruction Tuning for Facial Expression Quality Assessment25 Jun 2026
- 2026FlowCodec: One-Step Flow Prior for Generative Image Compression19 Jun 2026
- 2026NeoLoc-68: End-to-end 68-point neonatal facial landmark localisation in neonatal clinical environments18 Jun 2026
- 2026InfantFace: Detecting infant faces in neonatal clinical environments18 Jun 2026
- 2026MuVAP: Multimodal Multiparty Voice Activity Projection for Turn-taking Prediction in the Wild15 Jun 2026
- 2026BMD-45: A Large-Scale CCTV Vehicle Detection Dataset for Urban Traffic in Developing Cities27 Apr 2026
- 2026ActFER: Agentic Facial Expression Recognition via Active Tool-Augmented Visual Reasoning10 Apr 2026
- 2026IDDM: Identity-Decoupled Personalized Diffusion Models with a Tunable Privacy-Utility Trade-off1 Apr 2026
- 2025Knowledge Completes the Vision: A Multimodal Entity-aware Retrieval-Augmented Generation Framework for News Image Captioning26 Nov 2025
- 2025Photo Dating by Facial Age Aggregation7 Nov 2025
- 2025The Urban Vision Hackathon Dataset and Models: Towards Image Annotations and Accurate Vision Models for Indian Traffic4 Nov 2025
- 2025Identity-Preserving Image-to-Video Generation via Reward-Guided Optimization16 Oct 2025
- 2025Virtually Being: Customizing Camera-Controllable Video Diffusion Models with Multi-View Performance Captures16 Oct 2025
- 2025ContextGen: Contextual Layout Anchoring for Identity-Consistent Multi-Instance Generation13 Oct 2025
- 2025FakeParts: a New Family of AI-Generated DeepFakes28 Aug 2025
- 2025Age-Diverse Deepfake Dataset: Bridging the Age Gap in Deepfake Detection6 Aug 2025
- 2025Assessing the Use of Face Swapping Methods as Face Anonymizers in Videos27 May 2025
- 2025Automated Visual Attention Detection using Mobile Eye Tracking in Behavioral Classroom Studies12 May 2025
- 2025Transforming faces into video stories -- VideoFace2.04 May 2025
- 2025Exploring Cognitive and Aesthetic Causality for Multimodal Aspect-Based Sentiment Analysis22 Apr 2025
- 2025Insert Anything: Image Insertion via In-Context Editing in DiT21 Apr 2025
- 2025Trajectory-guided Motion Perception for Facial Expression Quality Assessment in Neurological Disorders13 Apr 2025
- 2025OmniTalker: One-shot Real-time Text-Driven Talking Audio-Video Generation With Multimodal Style Mimicking3 Apr 2025
- 2025Archival Faces: Detection of Faces in Digitized Historical Documents1 Apr 2025
- 2025Beyond Wide-Angle Images: Structure-to-Detail Video Portrait Correction via Unsupervised Spatiotemporal Adaptation1 Apr 2025
- 2025ComfyGPT: A Self-Optimizing Multi-Agent System for Comprehensive ComfyUI Workflow Generation22 Mar 2025
- 2025VideoGen-of-Thought: Step-by-step generating multi-shot video with minimal manual intervention19 Mar 2025
- 2025Concat-ID: Towards Universal Identity-Preserving Video Synthesis18 Mar 2025
- 2025Long-Term TalkingFace Generation via Motion-Prior Conditional Diffusion Model13 Feb 2025
- 2025Playmate: Flexible Control of Portrait Animation via 3D-Implicit Space Guided Diffusion11 Feb 2025
- 2025B-FPGM: Lightweight Face Detection via Bayesian-Optimized Soft FPGM Pruning28 Jan 2025
- 2025Eye Sclera for Fair Face Image Quality Assessment13 Jan 2025
- 2024VideoGen-of-Thought: Step-by-step generating multi-shot video with minimal manual intervention3 Dec 2024
- 2024CLIP Unreasonable Potential in Single-Shot Face Recognition19 Nov 2024
- 2024Template-based Multi-Domain Face Recognition15 Sep 2024
- 2024Towards Unified Facial Action Unit Recognition Framework by Large Language Models13 Sep 2024
50 newest of 86 All 86 (JSON)
- 2023DataComp: In search of the next generation of multimodal datasetscited by 635
- 2023GANDiffFace: Controllable Generation of Synthetic Datasets for Face Recognition with Realistic Variationscited by 49
- 2025Long-Term TalkingFace Generation via Motion-Prior Conditional Diffusion Modelcited by 49
- 2025ContextGen: Contextual Layout Anchoring for Identity-Consistent Multi-Instance Generationcited by 17
- 2024RectifID: Personalizing Rectified Flow with Anchored Classifier Guidancecited by 11
- 2025Playmate: Flexible Control of Portrait Animation via 3D-Implicit Space Guided Diffusioncited by 8
- 2024UVMap-ID: A Controllable and Personalized UV Map Generative Modelcited by 7
- 2025Identity-Preserving Image-to-Video Generation via Reward-Guided Optimizationcited by 7
- 2025Virtually Being: Customizing Camera-Controllable Video Diffusion Models with Multi-View Performance Capturescited by 4
- 2026IDDM: Identity-Decoupled Personalized Diffusion Models with a Tunable Privacy-Utility Trade-offcited by 1
- 2026FlowCodec: One-Step Flow Prior for Generative Image Compressioncited by 0
- 2026Diff-ID: Identity Consistent Facial Image Generation and Morphing via Diffusion Modelscited by 0
12 most cited of 13 All 13 (JSON)
- 2026ControlRef: Efficient Layout-Guided Multi-Instance Generation via Anchored 4D-RoPE7 Aug 2026
- 2026Diff-ID: Identity Consistent Facial Image Generation and Morphing via Diffusion Models27 Jul 2026
- 2026FlowCodec: One-Step Flow Prior for Generative Image Compression19 Jun 2026
- 2026IDDM: Identity-Decoupled Personalized Diffusion Models with a Tunable Privacy-Utility Trade-off1 Apr 2026
- 2025Identity-Preserving Image-to-Video Generation via Reward-Guided Optimization16 Oct 2025
- 2025Virtually Being: Customizing Camera-Controllable Video Diffusion Models with Multi-View Performance Captures16 Oct 2025
- 2025ContextGen: Contextual Layout Anchoring for Identity-Consistent Multi-Instance Generation13 Oct 2025
- 2025Long-Term TalkingFace Generation via Motion-Prior Conditional Diffusion Model13 Feb 2025
- 2025Playmate: Flexible Control of Portrait Animation via 3D-Implicit Space Guided Diffusion11 Feb 2025
- 2024RectifID: Personalizing Rectified Flow with Anchored Classifier Guidance23 May 2024
- 2024UVMap-ID: A Controllable and Personalized UV Map Generative Model22 Apr 2024
- 2023GANDiffFace: Controllable Generation of Synthetic Datasets for Face Recognition with Realistic Variations31 May 2023
12 newest of 13 All 13 (JSON)
- 2026Towards Spatial Supersensing in the Wildcited by 2
- 2026Towards Spatial Supersensing in the Wild15 Jul 2026
- 2022Expanding Accurate Person Recognition to New Altitudes and Ranges: The BRIAR Datasetcited by 61
- 2025Insert Anything: Image Insertion via In-Context Editing in DiTcited by 61
- 2021YOLO5Face: Why Reinventing a Face Detectorcited by 44
- 2025VideoGen-of-Thought: Step-by-step generating multi-shot video with minimal manual interventioncited by 33
- 2025Concat-ID: Towards Universal Identity-Preserving Video Synthesiscited by 28
- 2022There is a Time and Place for Reasoning Beyond the Imagecited by 22
- 2023Towards Realistic Generative 3D Face Modelscited by 20
- 2023Boosting Object Detection with Zero-Shot Day-Night Domain Adaptationcited by 17
- 20223DMM-RF: Convolutional Radiance Fields for 3D Face Modelingcited by 10
- 2025OmniTalker: One-shot Real-time Text-Driven Talking Audio-Video Generation With Multimodal Style Mimickingcited by 9
- 2025Exploring Cognitive and Aesthetic Causality for Multimodal Aspect-Based Sentiment Analysiscited by 6
- 2022EResFD: Rediscovery of the Effectiveness of Standard Convolution for Lightweight Face Detectioncited by 5
12 most cited of 64 All 64 (JSON)
- 2026Expression-Diverse References for Identity-Preserving Video Generation8 Oct 2026
- 2026Identity-Centric Video Summarization via Hierarchical Fusion of Biometric, Appearance, and 3D Body Features22 Sep 2026
- 2026Data Diversity, Not Frequency Invariance: A Controlled and Self-Audited Study of Compression-Robust Deepfake Detection26 Aug 2026
- 2026Localize, Don't Beautify: Client-Side Control of Image-Editing APIs for Cosmetic Surgery Previews3 Aug 2026
- 2026ID-V2V: Identity-Preserving Video Restylization24 Jul 2026
- 2026Articulated Humanoid Head for a Robot Receptionist Capable of Natural Human Interaction19 Jul 2026
- 2026DriveFace: A Cross-Spectral Through-Glass Face Dataset for On-the-Move Vehicular Border Control15 Jul 2026
- 2026Swapping Faces, Saving Features: A Dual-Purpose Pipeline for Pedestrian Privacy in ITS9 Jul 2026
- 2026VSRo-200: A Romanian Visual Speech Recognition Dataset for Studying Supervision and Multimodal Robustness9 Jul 2026
- 2026Optimizing Image Preparation and Compression for Face Recognition within 1024 Bytes29 Jun 2026
- 2026TraMP-LLaMA: Generative Interpretability with Decoupled Instruction Tuning for Facial Expression Quality Assessment25 Jun 2026
- 2026NeoLoc-68: End-to-end 68-point neonatal facial landmark localisation in neonatal clinical environments18 Jun 2026
12 newest of 64 All 64 (JSON)