Jaeyoung Kim

AI researcher

About Me

안녕하세요. 저는 AI researcher로 활동하고 있는 김재영이라고 합니다. 약 5년 동안의 실무 경험동안 다양한 태스크에 대한 딥러닝 모델 개발 및 연구들을 진행하고 있습니다. 의료 영상 도메인에서 데이터 설계부터 상용화 모델 개발까지의 경험과, 자연어 처리 분야에서 AI 면접 자동화 모델을 개발한 경험이 있으며 특정 도메인에 국한되지 않고 도전적인 문제들을 풀어나가는 것에 관심이 있습니다 :)

Skills

DL framework: Pytorch (2020~), Tensorflow (2018~2020)

Domain: Medical Imaging, Natural Language Processing

Programming: Python

Projects

VUNO Med-Fundus AI

VUNO Med-Fundus AI 모델 고도화 및 연구 리딩

(2021.09~2022.08)

Main Task Image classification.

Models CNN-based classifier, vessel segmentation model, optic & fovea segmentation model, and ungradable image classifier.

Modality Color fundus photography.

Approaches Semi-supervised Learning, Ensemble Learning, etc.

Result 마이그레이션 (Keras -> Torch) 및 제품 성능 향상

VUNO Med-Fundus Quantification

Hard Exuadate (HE) 정량화 모델 개발

(2022.02~2022.03)

Main Task Semantic segmentation.

Models CNN-based HE classifier, ROI segmentation model, and HE segmentation model.

Modality Color fundus photography.

Approaches Transfer Learning

Result 상용화 예정

VUNO Med-Dental AI (Periapical X-ray)

Periapical X-ray를 이용한 dental caries, cyst 탐지 모델 개발

(2020.03~2020.05)

Main Task Object detection.

Models Pretrained classifier and RetinaNet.

Modality Dental periapical X-ray.

Approaches Transfer Learning, Pseudo Labeling

Result 기술이전

VUNO Med-Dental AI (Panoramic X-ray)

VUNO Med-Dental AI 연구 및 인허가

(2018.04~2020.02)

Main Task Image classification.

Models CNN-based classifier, ROI segmentation model, and Tooth detection model.

Modality Panoramic X-ray.

Approaches Active Learning

Result 기술이전

AI 면접 자동화 시스템

Behavioral Event Interview (BEI) 를 이용한 면접자 역량평가 모델 개발

(2020.08~2021.07)

Main Task Text classification.

Model Pratrained language model.

Modality BEI transcription.

Approaches Multi-task Learning

Result 모델 성능 고도화

Competition

MICCAI

(2018)

Rank 2nd place at MICCAI REFUGE competition.

Task 1 Glaucoma classification

Task 2 Optic disc/cup segmentation

Careers

VUNO Inc.

Fundus Team 파트장

(2021.09~현재)

GENESISLAB Inc.

Research Scientist

(2020.10~2021.08)

VUNO Inc.

선행연구 Dental 파트장

(2020.02~2020.10)

VUNO Inc.

영상 R&D 연구원

(2018.04~2020.02)

Education

Gachon University

Biomedical Engineering (BS), Industrial Engineering (MS)

2013 - 2021

Papers

Leveraging the Generalization Ability of Deep Convolutional Neural Networks for Improving Classifiers for Color Fundus Photographs

Applied Sciences

https://www.mdpi.com/2076-3417/11/2/591

A two-stage deep learning-based system for patent citation recommendation

Scientometrics

https://link.springer.com/article/10.1007/s11192-022-04301-0

Text classification using capsules

Neurocomputing

https://www.sciencedirect.com/science/article/abs/pii/S0925231219314092

Refuge challenge) A unified framework for evaluating automated methods for glaucoma assessment from fundus photographs

Medical Image Analysis

https://www.sciencedirect.com/science/article/abs/pii/S1361841519301100

Patent document clustering with deep embeddings

Scientometrics

https://link.springer.com/article/10.1007/s11192-020-03396-7

DeNTNet) Deep Neural Transfer Network for the detection of periodontal bone loss using panoramic dental radiographs

Scientific Reports

https://www.nature.com/articles/s41598-019-53758-2