top of page
94463_86250_735_edited.jpg

AI Research Engineer

I research visual intelligence technology that analyzes/understands various phenomena in Image/Video, synthesizes visual information, or creates new images.

Also, I develop applications that utilize Large Language Models (such tasks are generally called Prompt Engineering).

1. AI/ML Modeling 

2. Prompt Engineering

3. AI Vision OCR

4. Object Recognition

5. Image Resolution

MAIN: 자기소개

Main Tasks (2021~ing)

1. VrDU (Visual rich Document Understanding) Modeling

Key Information Extraction (KIE) prior research

- Research fields: E2E document parsing, Table extraction, Layout analysis, Continual learning, Joint Learning, Document VQA
- Develop new services (B2B & B2C): frameworks and models design, experimentation, analysis

2. Prompt Engineering

Applied Large Language Model (LLM)

 - Development of chat service using Retrieval-Augmented Generation
 - Development of Intelligence Document Processing  using OCR and LLM

3. OCR Modeling

Developing techniques to distinguish printed or handwritten text characters within images

- End2end OCR R&D (Detector-Recognizer as one model)
- Research and development of Multimodal/Multidomain Pretraining model for OCR
- Research and development of various commercial models for document understanding (classification, detection, image de-warping, layout analysis, image matching)
- Domain generalization, data balancing, semi-supervised learning, Few-shot learning

4. Pre-training for Language Model

Adapt Language Models to Domains and Tasks

- Building specific corpora for performing multimodal tasks
- Pre-train LM from scratch on domain-specific corpora
- Adapt LM by additional pre-training on domain-specific corpora

MAIN: 기술

Projects

( 2021~ )

Shinhan Bank (December, 2021)

Anomaly detection -
Malicious Review Image Detection

I developed a classification model to review review images posted on the “땡겨요” delivery app. The main purpose of this model is to determine whether the image contains malicious content.

Shinhan Bank (December, 2022)

Documen AI -
Information Extraction

I conducted research called “Document AI” with the goal of automating financial tasks. This project focuses on technology to accurately extract only the necessary information from unstructured documents. These technologies are of great help in processing unstructured transaction documents frequently used in financial operations.

Shinhan Bank (January, 2023~ )

Prompt Engineering - 
Business Support Chatbot

I developed a business support chatbot using my LLM. The main purpose of this chatbot is to accurately respond to work-related questions and quickly find the information you need. In this project, I designed a data chunking strategy to build a Vector DB, established a reranking system, and constructed a benchmark.

Shinhan Bank (November, 2021~ )

Computer Vision -
Evaluate partners

I have a master's degree in computer vision. Currently, I am working as an evaluator evaluating the performance of various computer vision technologies used in the financial industry. In these activities, I am evaluating various computer vision systems, including ATM user abnormal behavior detection systems, ID recognition programs, virtual human technology, OCR, and Document AI.

MAIN: 경험 및 교육

Working Experience

2021 ~ ing

AI/ML Researcher

Shinhan Bank

Research in a wide range of areas, including object detection, image deblurring, natural language processing, user experience, and more.

2020 ~ 2022

Academic Research Scientist

Korea University

Research on deep learning, machine learning, computer vision, and more.

2017 ~ 2018

CEO

PCOY

Management, SW development, fashion MD, and more.

2017 ~ 2018

IT Instructor

Comseba

IT Coding Academy

Lecture on coding of C, C ++, JAVA, data structure, algorithm, and more.

MAIN: 경험

Education

M.S. in Korea Univ.

2022

Korea Unversity

M.S. in Electronics, Electrical and Computer Engineering
Multimedia Information Laboratory

2020

Jeonbuk National University

B.S. in Computer Science and Engineering

MAIN: 교육

Technical Skills

AI/ML

1. Deep Learning

2. Computer Vision

3. Object Recognition

4. Image Segmentation

5. Image Restoration

Development


1. Prompt Engineering

2. Search Engine

3. Full-stack Develop

Lecture

1. Programming

2. Data Structure


3. Algorithm

4. Information Olympiad

5. Coding Test

MAIN: 특기

Publication

Int'l Journal (SCI)

1. S. Cho, J. Moon, J. Bae, J. Kang, S. Lee. "A Framework for Understanding Unstructured Financial Documents using RPA and Multimodal Approach," Electronics, 2023.02. (IF: 2.690)

2. H. Kim, C. Kim, H. Kim, S. Cho, E. Hwang. “Panoptic Blind Image Inpainting,” ISA Transactions, 2023.02. (IF: 5.468)

3. S. Park, J. Moon, S. Cho, E. Hwang. “Instance Segmentation-based Review Photo Validation Scheme,” Journal of Supercomputing (JoS), 2022.08. (IF: 2.600)


4. H. Kim, H. Kim, S. Cho, E. Hwang. “An End-to-End Face Parsing Model Using Channel and Spatial Attentions,” Measurement, 2022.01. (IF: 5.131)​

MAIN: 텍스트

Int'l Conference

1. H. Kim, H. Kim, S. Cho, E. Hwang. "Manipulating Neural Network Block for Robust Image Segmentation," 2022 IEEE International Conference on Big Data and Smart Computing (IEEE BigComp 2022), Bangkok, Thailand, 2022.01.

2. S. Cho, H. Kim, B. Ko, E. Hwang. "Review Photo Validation Scheme Based on Faster R-CNN and Triplet Loss," The 6th International Conference on Next Generation Computing 2020 (ICNGC 2020), Busan, Korea, 2020.12. (Best Paper Award)

3. H. Kim, H. Kim, S. Cho, E. Hwang. "Attention Mechanism for Improving Facial Landmark Semantic Segmentation," The 22nd International Conference on Artificial Intelligence (ICAI 2020), Las Vegas, USA, 2020.07.

MAIN: 텍스트

Dom. Journal (KCI)

1. 조성국, 김형준, 정원용, 황인준. "지루성 두피염 진단을 위한 Faster R-CNN과 Atrous 컨볼루션 기반의 두피 각질 검출 기법". 정보과학회 컴퓨팅의 실제 논문지(KTCP), Vol. 27, No. 9, pp. 440-445, 2021.09.

2. 조성국, 김형준, 박성우, 황인준. "PP-YOLO를 이용한 실시간 두피 각질 검출 기법", 한국정보과학회 데이터베이스 연구회지, Vol. 37, No. 2, pp. 52-64, 2021.08.

MAIN: 텍스트

Dom. Conference

1. 조성국, 강지원, 이상욱, 오치훈, 김민수, "딥러닝과 데이터 증강을 활용한 악성 배달 음식 리뷰 이미지 탐지 기법", 2022 한국디지털콘텐츠학회 하계종합학술대회 (DCS 2022). 2022.07. (우수논문상)

2. 고범연, 김현우, 조성국, 황인준. "초해상화 GAN을 활용한 저해상도 얼굴 인식 기법", 한국정보과학회 2021 한국소프트웨어종합학술대회 (KSC 2021). 2021. 12.

3. 고범연, 김현우, 조성국, 황인준. "얼굴 특징점을 활용한 딥러닝 기반의 유사 얼굴 인식 기법", Korean DataBase Conference 2021 (KDBC 2021), 2021.11. (우수논문상)

4. 조성국, 김형준, 이지은, 황인준. "Faster R-CNN을 이용한 지루성 두피염 진단", 한국정보과학회 2020 한국소프트웨어종합학술대회 (KSC 2020), 2020.12. (우수발표논문상)

5. 조성국, 김형준, 황인준. "인스턴스 분할을 이용한 포토 리뷰 적합성 평가 기법", 한국정보과학회 2020 한국컴퓨터종합학술대회 (KCC 2020), 2020.07.

MAIN: 텍스트

AWARDS

1. 금상

2014 RGC 전국 로봇 페스티벌 (로봇 댄스 부문)

2. 대상

2019 로컬 소셜 이노베이션: PT 경진대회

3. 대상

2019 캡스톤 디자인 경진대회(작품명: AI와 CPS를 이용한 서빙 시스템)

4. 동상

2019 캡스톤 디자인 UCC 경진대회

5. 금상

2019 컴퓨터공학부 작품경진대회

6. Excellent Paper Award

The 6th International Conference on Next Generation Computing(Paper: Review Photo Validation Scheme Based on Faster R-CNN and Triplet Loss)

7. 우수발표논문상

한국정보과학회 2020 한국소프트웨어종합학술대회(논문명: Faster R-CNN을 이용한 지루성 두피염 진단)

8. 우수상

2021 산업융합 아이디어 사업화 해커톤(아이디어: 딥러닝 기반의 두피 질병 모니터링 시스템)

9. 우수논문상 대상

Korean DataBase Conference 2021(논문명: 얼굴 특징점을 활용한 딥러닝 기반의 유사 얼굴 인식 기법)

10. 우수논문상

한국디지털콘텐츠학회 2022 하계종합학술대회(논문명: 딥러닝과 데이터 증강을 활용한 악성 음식 리뷰 이미지 탐지 기법)

MAIN: 텍스트
bottom of page