Face Detection, Recognition and Liveness
Finding faces, verifying identity, and checking attributes, gaze or spoofing for identity and safety use.
Tier Grades
Tier A: Frontier AI labs
feasible
Academia and industry labs set face recognition methods such as ArcFace.
Tier B: Indonesian AI labs
feasible
Training face models on local faces is within an Indonesian lab reach, subject to consent and data law.
Tier C: Institution units such as ITB
feasible
A 6-author paper set the method; a university unit can work on it.
Tier D: Organisations such as SEMAI (PoT, PoC, PoV)
feasible
Released face and liveness models support eKYC and access pilots, with privacy review.
Where this comes from
Tiers: a) frontier AI labs, b) Indonesian AI labs, c) institution units such as ITB, d) organisations such as SEMAI delivering proof of technology, concept and value. Each rationale cites the anchors below.
Anchors
- ComputeTraining compute is not in the sources read.
- DataNot in the sources read; ArcFace addresses massive label noise in raw web faces.
- Team size (proxy)Proxy only (author count of an exemplar paper, not a headcount): 6 authors on the ArcFace paper (arXiv 1801.07698).
- Time horizonYears.
- Shelf-life signalsRegulation tightens use of face recognition, Deepfake attacks force new liveness checks
- Pipeline roleTier A and academia publish methods; Tier B and Tier C test them on Indonesian faces; Tier D builds eKYC and driver monitoring pilots.
Where this comes from
Team size is a proxy (author counts of exemplar papers), not a headcount.
Who and What
Exemplars
Observed actors
Linked models
BlazeFacevisionexemplar
RetinaFacevisionexemplarArcFacevisionexemplarFace anti-spoofing MobileNetV3visionexemplar6DRepNetvisionproduced bySCRFDvisionproduced by
YOLOv5-Facevisionproduced byMediaPipe Irisvisionproduced byCavaFacevisionproduced byFace Meshvisionproduced byMobileFaceNetvisionproduced by
Emotion FERPlusvisionproduced byMTCNNvisionproduced byFaceNetvisionproduced byMediaPipe vision modelsvisionproduced byUltra-Light-Fast-Generic-Face-Detector (UltraFace)visionproduced by
FaceBoxesvisionproduced byYuNetvisionproduced by
DamoFDvisionproduced byMogFacevisionproduced byInspireFacevisionproduced byL2CS-Netvisionproduced by
Intel face modelsvisionproduced by
FaceDetect and FaceDetectIRvisionproduced byLinks
R&D TopicFace Detection, Recognition and Liveness
ModelsBlazeFaceGoogle, visionArcFaceInsightFace (Deep Insight), visionRetinaFaceInsightFace (Deep Insight), visionFace anti-spoofing MobileNetV3vision+25 more
BlazeFaceGoogle, visionArcFaceInsightFace (Deep Insight), visionRetinaFaceInsightFace (Deep Insight), visionFace anti-spoofing MobileNetV3vision+25 moreUse Casesnone recorded
Who Runs Itnone recorded
Models 29
MediaPipe vision modelsGoogle, visionproduced byBlazeFaceGoogle, visionexemplarFace MeshGoogle, visionproduced byDamoFDAlibaba DAMO, visionproduced byFaceDetect and FaceDetectIRNVIDIA, visionproduced byIntel face modelsIntel, visionproduced byMediaPipe IrisGoogle, visionproduced by
Qualcomm face modelsQualcomm, visionproduced bySCRFDInsightFace (Deep Insight), visionproduced byArcFaceInsightFace (Deep Insight), visionexemplarEmotion FERPlusMicrosoft Research, visionproduced byFaceBoxesInstitute of Automation, Chinese Academy of Sciences, visionproduced by+17
FaceNetGoogle, visionproduced byInspireFacevisionproduced byMobileFaceNetvisionproduced byMogFaceAlibaba DAMO, visionproduced byRetinaFaceInsightFace (Deep Insight), visionexemplarUltra-Light-Fast-Generic-Face-Detector (UltraFace)visionproduced byYOLOv5-FaceShenzhen Deepcam Information Technologies, visionproduced byYuNetvisionproduced by6DRepNetvisionproduced byAge and Gender Classification CNNvisionproduced byCavaFacevisionproduced byEyeGazevisionproduced byFace anti-spoofing MobileNetV3visionexemplarL2CS-Netvisionproduced byMTCNNvisionproduced byOpen-closed-eye-0001visionproduced byWHENetvisionproduced byPapers
- ArcFace: Additive Angular Margin Loss for Deep Face RecognitionJiankang Deng et al., arXiv preprint, 2018 founding
- Joint Face Detection and Alignment using Multi-task Cascaded Convolutional NetworksKaipeng Zhang et al., arXiv preprint, 2016 founding
- AdaFace: Quality Adaptive Margin for Face RecognitionMinchul Kim et al., CVPR2022 (per arXiv comment), 2022 state of the art
- RetinaFace: Single-stage Dense Face Localisation in the WildJiankang Deng et al., arXiv preprint, 2019 state of the art
- Deep Learning for Face Anti-Spoofing: A SurveyZitong Yu et al., arXiv preprint, 2021 survey
Where this comes from
The papers to read first: where the direction began, its current best, and a survey when one exists.
Sources
- arxiv.org/abs/1801.07698 https://arxiv.org/abs/1801.07698
To cite this page: Project AURION, AI Use-Cases Research & Innovation, "Face Detection, Recognition and Liveness", https://aurion.semai.id/topics/face-analysis-and-liveness/, records of October 2026.