ChanMin Park

Vision Research Engineer

I build industrial and medical vision systems end to end: from Physical AI and domain adaptation to 3D detection, MLOps, and production deployment.

Physical AI · Spatial Intelligence

Build Physical AI, sim-to-real, and autonomous vision systems for manufacturing processes.

Physical AI Spatial Intelligence Vision Control

System Engineering · MLOps

Design end-to-end vision systems from data and model pipelines to packaging, deployment, and reliable operations.

System Engineering MLOps Deployment

Sim2Real · SSL · Domain Adaptation

Develop sim-to-real, self-supervised learning, and domain adaptation workflows that keep vision models stable under domain shift.

Sim2Real Self-Supervised Domain Adaptation

📞 010-3999-3403 ✉️ Email 💼 LinkedIn 💻 GitHub 📄 CV

5+ Years in CV & Industry

5+ Major projects

5+ Patents

6 Vision papers

Experience

AX Technology R&D Group · Spatial Intelligence Vision Engineer · POSCO DX

Pangyo, Korea · July 2024 - Present

Physical AI Isaac Sim PLC Integration Dataiku MLOps Image Retrieval

Isaac Sim·PLC·Sim2Real: Built physical-AI training environments that mirrored steel-site conditions and iterated models against real deployment constraints.
Dataiku·SDD·MLOps: Advanced Digital Transformation (DX) from smart factory to intelligence factory by building SDD-driven image retrieval and automated MLOps pipelines for steel vision systems.
Autonomous Steelmaking·L2 Integration·Vision AI: Integrated L2 communications and control signals to apply vision-based anomaly detection in autonomous steelmaking operations.

사용한 tools:

Isaac Sim

PyTorch

Dataiku

Docker

MLflow

AI Research Engineer · VUNO Inc.

Seoul, Korea · May 2021 - July 2024

Medical AI PyTorch Domain Adaptation Self-Supervised Learning 3D Detection Applied Engineering MLflow

Test-Time Adaptation·Domain Adaptation·ACCV 2024: Developed adaptation strategies for medical imaging domain shifts and contributed to the CNG-SFDA research line published at ACCV 2024.
Self-Supervised Learning·Universal Segmentation·TIGER Challenge: Conducted segmentation-oriented SSL research, built pathology workflows around it, and applied the stack to achieve 1st place in the TIGER Challenge.
Lung CT·3D Tiny Object Detection·MLOps: Built and deployed lung nodule detection pipelines for 3D tiny-object modeling, with production-grade monitoring and release workflows in real service environments.
PathQuant·End-to-End Delivery·Applied Engineering: Owned the full cycle from data collection and model refinement to packaging and final deployment, demonstrating strong applied-engineer execution in medical imaging products.

사용한 tools:

PyTorch

Docker

MLflow

Education

M.S. in Computer Science · UNIST

Ulsan, Korea · Mar 2019 - Mar 2021

Cumulative GPA 3.28/4.3, Scholarships: Academic excellence for 4 semesters (2019-2021)

Thesis: Neuron segmentation using incomplete and noisy labels via adaptive learning with structure priors

Domestic Patent: Brain Neural Network Structure Image Processing System, Brain Neural Network Structure Image Processing Method, and a computer-readable storage medium

B.S. in Biomedical Engineering · Inje University

Gimhae, Korea · Mar 2014 - Aug 2018

Cumulative GPA 3.89/4.5, Scholarships: Academic excellence for 8 semesters (2014-2018)

Awards: Gold Prize: Inje Creative Comprehensive Design Contest; Bronze Prize: 2017 Engineering Festival Competition; Good Capstone Design Contest

Projects

Steel Defect Detection

[Detail]

Steel Defect Detection

PyTorch GAN Anomaly Detection

Pathology Detection

PyTorch Object Detection Whole Slide Image

Lung CT Detection

3D CNN Nodule Detection Medical AI

TIGER Challenge

Multi-Task Segmentation Grand Challenge

MRI Analysis

MRI Registration Brain Analysis

Segmentation

U-Net Semantic Seg Multi-Organ

Image Reconstruction

WGAN-GP Super Resolution CT/MRI

Test-Time Adaptation

Domain Adaptation CNG-SFDA Self-Training

Video Action

Video Action Recognition Temporal

Seminar & Teaching

딥러닝을 활용한 의료 영상 처리 & 모델 개발

Part-Time Instructor · FastCampus Inc.

Seoul, Korea · Sep 2023 - Dec 2023

[Online Course]

의용공학과에서 AI연구원되기까지

Career Mentor · Inje University

Gimhae, Korea · 2023 - 2024

Provided guidance and mentorship to students in computer vision and machine learning projects.

Image Classification · Machine Learning Course

Instructor · Gachon University (VUNO)

Seongnam, Korea · 2023 - 2024

Taught machine learning fundamentals and applications to university students.

Publications & Research

CNG-SFDA: Clean-and-Noisy Region Guided Online-Offline Source-Free Domain Adaptation

Cho, H., Park, C., Kim, D. H., Kim, J., & Kim, W. H.

ACCV 2024

[PDF]

Joint-Embedding Predictive Architecture for Self-Supervised Learning of Mask Classification Architecture

Dong, H., Jo, D., Cho, H., Park, C., & Kim, W. H.

arXiv, 2024

[PDF]

Neuron segmentation using incomplete and noisy labels via adaptive learning with structure priors

Park, Chanmin, et al.

ISBI 2021

[PDF]

Tumor-infiltrating lymphocytes in breast cancer through artificial intelligence: biomarker analysis from the results of the TIGER challenge

van Rijthoven, M., ... Park, C., ... & Ciompi, F.

medRxiv, 2025

[PDF]

데이터 기반 동적 필터링이 포인트 클라우드 객체 증분 학습에 미치는 영향도 분석

진종현, 박찬민, 윤일용

한국정보과학회 학술발표논문집, 2025

Development of a KFDA-certified Deep Learning Algorithm for Quantitative Analysis of Ki-67 Immunohistochemical Stains

Park, C., et al.

대한병리학회 (Domestic Conference)

Patents

Method for analyzing medical image (Application number: 1020210139898)
Method For Analyzing Medical Image based on deep learning (Application number: 1020210139897)
Brain Neural Network Structure Image Processing System, Brain Neural Network Structure Image Processing Method, and a computer-readable storage medium (Registration number: 1020210139898)

Expertise & Tools

🎯

Object Detection & Segmentation

📦

3D Vision · Depth · Pose

🤖

Sensor Fusion & Robotics Perception

🎥

Video Recognition & Tracking

🧬

Self-Supervised Vision (DINO/MoCo/MAE)

🕸️

Embedding & Metric Learning

🔄

Domain Adaptation & Continual Eval

💾

Large-Scale Data Pipelines

🚀

Edge Deployment & MLOps

Python

PyTorch

TensorFlow

Docker

Linux

MongoDB

AWS

MLflow

Dataiku

ROS2

Isaac Sim

Git

Confluence

Jira

Awards & Competitions

🥇 TIGER Challenge - 1st PlaceVUNO
🥇 HealthHub Datathon - 1st Place2020
🏆 Video Recognition: Google - Isolated Sign Language Recognition
🥇 Instance Segmentation: HealthHub Datathon Track B - 1st Place2020
🎯 Object Detection: Lesion Detection AI Competition Code Link
🏅 Tabular Classification: Autonomous Driving Sensor - Top 8%
Inje Creative Comprehensive Design Contest🥇 Gold
2017 Engineering Festival Competition🥉 Bronze
Good Capstone Design Contest🏅 Excellence

Experience

AX Technology R&D Group · Spatial Intelligence Vision Engineer · POSCO DX

AI Research Engineer · VUNO Inc.

Education

M.S. in Computer Science · UNIST

B.S. in Biomedical Engineering · Inje University

Projects

Steel Defect Detection

Seminar & Teaching

딥러닝을 활용한 의료 영상 처리 & 모델 개발

의용공학과에서 AI연구원되기까지

Image Classification · Machine Learning Course

Publications & Research

CNG-SFDA: Clean-and-Noisy Region Guided Online-Offline Source-Free Domain Adaptation

Joint-Embedding Predictive Architecture for Self-Supervised Learning of Mask Classification Architecture

Neuron segmentation using incomplete and noisy labels via adaptive learning with structure priors

Tumor-infiltrating lymphocytes in breast cancer through artificial intelligence: biomarker analysis from the results of the TIGER challenge

데이터 기반 동적 필터링이 포인트 클라우드 객체 증분 학습에 미치는 영향도 분석

Development of a KFDA-certified Deep Learning Algorithm for Quantitative Analysis of Ki-67 Immunohistochemical Stains

Patents

Expertise & Tools

Awards & Competitions

Recent Posts

[PaperReview] UCDR-Adapter: Exploring Adaptation of Pre-Trained Vision-Language Models for Universal Cross-Domain Retrieval

[PaperReview] MaPLe: Multi-modal Prompt Learning

[PaperReview] DePro: Domain Ensemble using Decoupled Prompts for Universal Cross-Domain Retrieval

[PaperReview] ProS: Prompting-to-Simulate Generalized Knowledge for Universal Cross-Domain Retrieval

[ROS2] 01: Introduction (ROS2 소개)

[PaperReview] Semantic Feature Learning for Universal Unsupervised Cross-Domain Retrieval

[PaperReview] Unsupervised Cross-domain Image Retrieval via Prototypical Optimal Transport

[PaperReview] Noise Mitigation for Unsupervised Cross-Domain Image Retrieval

[PaperReview] Feature Representation Learning for Unsupervised Cross-domain Image Retrieval

[PaperReview] Unsupervised Feature Representation Learning for Domain-Generalized Cross-Domain Image Retrieval

[PaperReview] Correspondence-free Domain Alignment for Unsupervised Cross-domain Image Retrieval

[CS231A] Lecture 10: Optimal Estimation (최적 추정)

[CS231A] Lecture 09: Optical and Scene Flow (광학 및 장면 흐름)

[CS231A] Lecture 08: Monocular Depth Estimation (단안 깊이 추정)

[CS231A] Lecture 07: Representation Learning (표현 학습)

[CS231A] Lecture 06: Fitting and Matching (피팅 및 매칭)

[CS231A] Lecture 05: Active and Volumetric Stereo (능동 및 볼륨 스테레오)

[CS231A] Lecture 04: Stereo Systems (스테레오 시스템)

[CS231A] Lecture 03: Epipolar Geometry (에피폴라 기하학)

[CS231A] Lecture 02: Single View Metrology (단일 뷰 측정학)

[CS231A] Lecture 01: Camera Models (카메라 모델)

[CS231A] Lecture 02: Single View Metrology (단일 뷰 측정학)

3D Vision Tutorial

Image Transformation and Warping

Image Processing Basics

Advanced Image Filtering Techniques

Vision Transformer Primer

Deep Learning Systems for Reliable Vision

Physically Based Rendering Cheat Sheet

3D Geometry Fundamentals for Vision

[PaperReview] SAM 3D: 3Dfy Anything in Images