About Me

My goal is a self-evolving dark factory, where production lines keep improving on their own with little human intervention. I work toward this goal at Samsung Electronics by building industrial AI with foundation models. My current focus is Vision Foundation Models (VFM) and Embodied Recognition Models (ER) for visual inspection.

Education

  • Ph.D. in Electrical Engineering and Computer Science
    Gwangju Institute of Science and Technology (GIST)
    Mar 2016 – Feb 2022 · Advisor: Prof. Moongu Jeon

Work Experience

  1. Staff Engineer

    Jan 2026 – Present
    Samsung Electronics — AI Inspection Technology Group, Global Technology Research (GTR) Suwon, Korea
    Show moreShow less
    • Vision Foundation Model (VFM) research for manufacturing visual inspection (Smart Factory Vision).
    • DX AI Vision Platform and its data agent for deploying inspection models across production lines.
  2. Staff Engineer

    Mar 2022 – Dec 2025
    Samsung Electronics — Factory Automation Technology Group, Global Technology Research (GTR) Suwon, Korea
    Show moreShow less
    • Foundation-model-based vision inspection technology (Samsung IT Innovation Awards Grand Prize, 2024).

Publications

BibTeX

Showing all 24 publications

2027

Not All Lines Are Equal: Input-Adaptive Line Segment Detection

Younkwan Lee, Jueun Ko

IEEE/CVF Winter Conference on Applications of Computer Vision (WACV), 2027.

To appear

2026

2025

2024

2022

2021

2020

2019

Other international conference papers4
  1. Empirical Analysis of Tree-Based Models for PM2.5 Concentration Prediction Juhyun Lee, Yoojin Hong, Younkwan Lee, Hyun Soo Kim, Chul Han Song, Du Yong Kim, Moongu Jeon International Conference on Signal Processing and Communication Systems (ICSPCS), pp. 1–7, 2019.
  2. Accurate License Plate Recognition and Super-Resolution Using Generative Adversarial Networks on Traffic Surveillance Video Younkwan Lee, Jaewoong Yun, Yoojin Hong, Juhyun Lee, Moongu Jeon IEEE International Conference on Consumer Electronics – Asia (ICCE-Asia), pp. 1–4, 2018.
  3. Real-Time Traffic Sign Recognition Quang Vinh Dinh, Younkwan Lee, Hyunguk Choi, Moongu Jeon IEEE International Conference on Consumer Electronics – Asia (ICCE-Asia), 2018.
  4. Automatic Part Localization Using 3D Cuboid Box for Vehicle Subcategory Recognition Younkwan Lee, Jongmin Yu, Moongu Jeon International Conference on Control, Automation and Information Sciences (ICCAIS), pp. 175–180, 2017.
Domestic conference papers (Korea)9
  1. Comparative Analysis of High-Level Vision Task Performance under Image Denoising (Image Denoising에 따른 High-Level Vision Task 성능 비교 분석) Younkwan Lee, Jihyo Jeon, Daeyoung Han, Moongu Jeon Korea Computer Congress (KCC), Korean Institute of Information Scientists and Engineers, 2020.
  2. Attention-based Self-Supervised Monocular Depth Estimation (어텐션 기반 자기지도학습 단안 영상 깊이 추정) Jihyo Jeon, Younkwan Lee, Daeyoung Han, Moongu Jeon Korea Computer Congress (KCC), Korean Institute of Information Scientists and Engineers, pp. 779–781, 2020.
  3. Analysis of Data Resolution in Real-World Image Classification (해상도에 따른 실세계 이미지 분류 고찰) Hyeongjun Yoo, Younkwan Lee, Moongu Jeon Korea Computer Congress (KCC), Korean Institute of Information Scientists and Engineers, pp. 729–731, 2020.
  4. Learning from Imbalanced Data with Cross Entropy (Cross Entropy를 이용한 불균형 데이터의 학습) Yechan Kim, Younkwan Lee, Moongu Jeon Korea Computer Congress (KCC), Korean Institute of Information Scientists and Engineers, 2020. Excellence Paper Prize
  5. Self-Supervised Monocular Depth Estimation (자기지도적 학습에 의한 단안 영상 깊이 추정) Jihyo Jeon, Younkwan Lee, Juhyun Lee, Yeongmin Ko, Moongu Jeon Workshop on Image Processing and Image Understanding (IPIU), 2020.
  6. Rotation-Invariant Self-Supervised Object Detection (회전 불변의 자기지도적 물체 검출) Younkwan Lee, Juhyun Lee, Jiwon Jun, Yeongmin Song, Moongu Jeon Korea Software Congress (KSC), Korean Institute of Information Scientists and Engineers, 2019.
  7. Empirical Analysis of Machine-Learning-Based Missing Value Imputation for PM10 Prediction (PM10 농도 예측을 위한 머신러닝 기반 결측치 처리의 실증적 분석) Juhyun Lee, Younkwan Lee, Yoojin Hong, Moongu Jeon Summer Annual Conference of the Institute of Electronics and Information Engineers (IEIE), 2019.
  8. A CNN-Based Single-Pipeline Detection Method for License Plate Detection in Unconstrained Environments (비정형화된 환경에서 차량 번호판 검출을 위한 CNN 기반의 단일 파이프라인 검출 방법) Younkwan Lee, Yoojin Hong, Jaewoong Yun, Moongu Jeon Korea Software Congress (KSC), Korean Institute of Information Scientists and Engineers, pp. 910–912, 2017. Excellence Paper Prize
  9. Scene Text Detection Using a Text Synthesis Model (자연의 영상에서 문자 합성 모델을 사용한 문자 검출) Younkwan Lee, Dae-Young Shim, Moongu Jeon Summer Annual Conference of the Institute of Electronics and Information Engineers (IEIE), 2017.

Industry Projects

R&D programs I have worked on at Samsung Electronics, from vision inspection foundation models to the MLOps and robotics that put them on production lines.

Vision Foundation Model for AI Inspection

Apr 2026 – Dec 2026

Foundation model for manufacturing inspection that unifies external and in-house AI through grounding.

Show moreShow less
  • Co-training of the VFM with LLM/VLM components for agentic inspection workflows.
  • Selectable inference modes: VFM only, instruction-conditioned, and post-hoc explanation.
  • Inspection data curated into reusable AI assets for domain specificity.
Vision Foundation ModelGroundingVLMAgentic AI

Re-inspection Automation Linked to AI Inspection Results

Feb 2026 – Nov 2026

Robotic re-inspection that imitates how inspectors handle a product to confirm suspected defects.

Show moreShow less
  • Target observation poses and robot motions generated from AI inspection results.
  • Pairwise-ranking model that prioritizes viewing angles and positions per defect type.
  • Multi-DOF hardware module for product manipulation under variable lighting and viewpoints.
RoboticsActive VisionPairwise RankingSmart Factory

DX AI Vision Platform Advancement

Dec 2025 – Oct 2026

Extending the AI vision inspection platform into the DX standard platform with cloud services and AI agents.

Show moreShow less
  • Cloud-based infrastructure for large-scale inspection data and data-driven decision support.
  • AI agent that makes the VFM usable by field engineers (Data Agent, group Annual Award 2026).
  • Extension to external sites such as package-damage inspection at logistics centers, with per-site MLOps.
MLOpsCloudAI AgentData Platform

Cloud-based Inspection Quality AI Service

Dec 2024 – Dec 2025

Cloud MLOps service for deploying and maintaining deep-learning vision inspection in the field.

Show moreShow less
  • Cloud-native AI service built on a load balancer, an MLOps platform, and a Kubernetes engine.
  • Vision–text foundation model and generative AI for inspection.
  • Few-shot customization of inspection models from small data (industry–academia collaboration).
MLOpsKubernetesVision–Language ModelFew-shot

MLOps AI Technology for Intelligent Inspection

Mar 2023 – Jan 2024

MLOps solution that keeps product-inspection models performant with minimal manual effort.

Show moreShow less
  • Data cleansing and automatic discovery of new defect types to cut data-management effort.
  • Product-specific model generation and lifelong learning to shorten training and validation cycles.
  • Open-set model updates and edge deployment and monitoring with less engineer intervention.
MLOpsLifelong LearningOpen-set RecognitionEdge Deployment

TV Exterior Inspection System

Jul 2022 – Nov 2022

Deep-learning exterior inspection for large-size TVs to reduce cosmetic defects.

Show moreShow less
  • Gap and step inspection system installed and verified on a pilot production line.
  • Rear-cover inspection with high-resolution tiling, candidate detection, and false-call removal by classification.
Defect DetectionDeep LearningProduction Deployment

Awards & Honors

  • 2026
    Annual Award, AI Inspection Technology Group
    Samsung Electronics · Oct 2026 — Development of the DX AI Vision Platform Data Agent
  • 2026
    Outstanding Reviewer
    IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR 2026) · Jun 2026 Certificate
  • 2025
    Samsung Best Paper Awards (SBPA) — Merit Award
    Samsung Electronics · 2025 · top 4.6% of submissions — HDO: Hierarchical Defect Ontology for Open-Vocabulary Defect Detection (Younkwan Lee, Byeongeun Ko, Eungi Hong)
  • 2024
    Samsung IT Innovation Awards (SIIA) — Grand Prize
    Samsung Group (group-wide competition) · Oct 2024 — Project representative, Vision Inspection Foundation Model
  • 2024
    Best Paper Runner-Up Award
    IEEE International Conference on Advanced Video and Signal Based Surveillance (AVSS 2024) · Jul 2024 — Self-Supervised Industrial Image Generation for Defect Recognition Under Limited Data Certificate
  • 2022
    Excellent Research Award (ETRI President's Award)
    Electronics and Telecommunications Research Institute (ETRI) · Feb 2022
  • 2021
    Excellence Paper Prize
    Korea Computer Congress (KCC 2021) · Jul 2021
  • 2020
    Best Student Paper Award
    IEEE Intelligent Vehicles Symposium (IV 2020) — Context-Aware Multi-Task Learning for Traffic Scene Recognition in Autonomous Vehicles
  • 2020
    Excellence Paper Prize
    Korea Computer Congress (KCC 2020) · Mar 2020
  • 2019
    Outstanding Paper Award
    RLQ Workshop at ICCV 2019 · Oct 2019 — SNIDER: Single Noisy Image Denoising and Rectification for Improving License Plate Recognition
  • 2019
    Outstanding Paper Award
    Gwangju Institute of Science and Technology (GIST) · Jan 2019
  • 2017
    Excellence Paper Prize
    Korea Software Congress (KSC 2017) · Dec 2017
  • 2016–2022
    Korean Government Scholarship (M.S./Ph.D.)
    Government of the Republic of Korea

Talks & Service

Invited Talks

  • Fast Object Detection in Smart Factory
    POSCO Technical Research Laboratories, Gwangyang, Korea · Nov 2019
  • Deep Learning Based License Plate Recognition
    DeepView Seminar, ETRI, Daejeon, Korea · May 2019
  • Smart Factory and Generative Adversarial Networks
    POSCO Technical Research Laboratories, Gwangyang, Korea · Aug 2018

Organizing

  • DeepView Workshop at IEEE AVSS 2021 — co-organizer

Reviewer

  • IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) — Outstanding Reviewer, 2026
  • IEEE Transactions on Intelligent Transportation Systems (T-ITS)
  • IEEE Transactions on Circuits and Systems for Video Technology (TCSVT)
  • IEEE Access
  • Information Sciences
  • IET Image Processing
  • IEEE Intelligent Vehicles Symposium (IV)

Teaching

  • Teaching Assistant, Machine Learning (EC4213), GIST, 2017
  • Teaching Assistant, Computer Graphics, GIST, 2015

Patents

  1. Vehicle License Plate Recognition SystemYounkwan Lee, Moongu Jeon · KR 10-2021-0001849 (filed Jan 2021)
  2. Access Control System Using Continual-Learning-Based Face RecognitionYounkwan Lee, Moongu Jeon · KR 10-2021-0001850 (filed Jan 2021)
  3. Rain Removal System for ImagesYounkwan Lee, Hyeongjun Yoo, Moongu Jeon · KR 10-2020-0108866 (filed Dec 2020)
  4. Apparatus and Method for Learning Video ImagesGeunshin Lee, Jongsun Lee, Moongu Jeon, Younkwan Lee · KR 10-2119687 (registered Jun 2020)
  5. Image Enhancement ApparatusGeunshin Lee, Jongsun Lee, Moongu Jeon, Younkwan Lee · KR 10-2124497 (registered Jun 2020)
  6. Action Recognition Method and ApparatusYongsang Yoon, Moongu Jeon, Jongmin Yu, Younkwan Lee · KR 10-2019-0123041 (filed Oct 2019)
  7. Vehicle Recognition SystemYounkwan Lee, Moongu Jeon · KR 10-2082254 (registered Feb 2020)