Skip to content
View seoyounji's full-sized avatar
  • DGIST
  • south korea
Block or Report

Block or report seoyounji

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Please don't include any personal information such as legal names or email addresses. Maximum 100 characters, markdown supported. This note will be visible to only you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
seoyounji/README.md

Hi there 👋

Hits

멋진 머신러닝 개발자가 되고 싶은 주니어 개발자입니다!

🇰🇷 Republic of Korea
🎂 Feb. 1996

 Anurag의 github 통계

Top Langs

Skills ⭐

기술명 설명 숙련도
Tensorflow1 Classification / Object Detection 을 사용해 웹앱 개발 경험 있음. 70%
Tensorflow2 논문을 읽고 YOLO 를 직접 구현해 본 적 있음. 30%
C++ OpenCV / OpenGL 을 이용해 실시간 3d Reconstruction 을 구현해 논문 제출 경험 있음. 70%
Python 딥러닝 모델을 읽고 사용 가능. 70%
Spring MVC & MVVM 패턴을 사용해 백엔드 구축 가능. 60%
Java 알고리즘 문제를 풀 때 주로 사용. 80%
Matlab 머신러닝 학습 데이터 생성 시 Data Preprocessing 경험 있음. 40%
Git&Github Git Branch를 사용해 협업 프로젝트 진행 경험 있음. 70%
MySQL 프로젝트 진행 시 사용해 DB를 구축함. 70%

Education 🎓

  • Gimpo Foreign Language High School (2011~2014)

  • Bachelor's degree in Computer Science and Engineering @ DGIST (2014~2018)

  • Master's degree in Computer Vision @ DGIST VC lab (2018~2020)

Experiences 📣

  • IPIU 2020 Paper Accept (1st Author) - 머리 부분 탑재형 디스플레이를 이용한 실시간 3차원 눈 표정 재건축

    Paper Abstract

    ​ While recently some approaches to capture facial expressions of users wearing HMDs have been proposed, most of them can handle only a limited set of eye expressions such as opening and closing of eyes. To overcome this, in this paper, we propose a novel approach to recognition and reconstruction of detailed facial expression of a user wearing a HMD in the eye region in real-time. Our approach models the eye expression of a user using a 3D blenshape model to enable 3D reconstruction of eye expression. To capture eye expression of a user, our approach adopts infra-red (IR) cameras installed inside a HMD, and a light-weight neutral network model that takes a pair of images of both eyes captured by IR cameras, and predicts the blendshape parameters. To train our neural network, we built a dataset by capturing real eye images of users and manually annotating them. Experimental results show that our approach can handle diverse eye expressions such as frowning compared to previous approaches.

  • SSAFY (Samsung Software Academy for Youth) 4th (2020~2021)

  • Participate SIGGRAPH 2019 / IPIU 2020, 2019, 2018 / ICCV 2019

Projects 🌈

  • 신세계 I&C 기업 연계 프로젝트 - 셀프스토어 활성화를 위한 가상 장바구니 App Web 사이트 개발 @SSAFY (진행중)

  • 시각 장애인을 위한 장보기 앱 See장 @SSAFY (2021)

    • SSAFY 공통 프로젝트 1등 / UCC 3등
    • Object Detection을 이용해 찾고자 하는 상품명을 말하면 상품이 사용자 위치를 중심으로 어디에 위치해 있는지 음성 출력으로 알려줌.
    • Image Classification을 이용해 손에 들고 있는 상품을 판별해 음성 출력으로 알려줌.
    • 맡은 역할 : Object detection, Image Classification 구현 전반 (학습용 데이터 생성 및 학습)
  • 부동산 매물 거래가 내역 확인을 위한 웹사이트 HappyHouse @SSAFY (2020)

    • SSAFY 관통 프로젝트 2등
    • 서울시 부동산 매물 거래 정보를 확인할 수 있고 지도를 통해 매물 정보 및 주변 상권과 편의시설 정보 또한 확인 가능.
    • 관심동네로 설정한 동네의 매물을 한눈에 볼 수 있고 주위에 원하는 상권이나 편의시설의 유무 또한 쉽게 확인 가능.
    • 관심동네로 설정하지 않은 다른 동네 중 원하는 상권이나 편의시설이 있는 동네의 매물을 추천해줌.
    • 맡은 역할 : DB 데이터 정제, 메인페이지 제작 및 전체 UI css form 제작, 매물 지도 표시 기능, 매물 검색 기능, 매물 상세보기 페이지 제작, 주변 상권정보 지도 표시 기능, 찜 목록 페이지 제작.
  • 초실감 원격가상 인터랙션을 위한 사용자 복원 기술 개발 프로젝트 @DGIST VC lab (2018~2020)

    • 한국 연구재단 지원
    • 참여 연구소 : 고려대학교, KAIST, POSTECH, DGIST
    • 카메라를 이용해 사용자의 모습을 촬영한 것을 3차원 컴퓨터 상에서 실시간으로 렌더링해 궁극적으로 서로 다른 장소에 있는 다수의 사람들이 렌더링된 서로를 만날 수 있게끔 하는 프로젝트.
    • 맡은 역할 : DGIST 연구 책임자로써 얼굴 표정 중 눈 부분을 집중적으로 모델링.
    • 3d Reconstruction 방식 중 Blendshape 방식을 이용해 진행하였으며 이 과정에서 눈 표정에 특화된 parameter 를 자체적으로 만들고 성공적으로 학습시켜 실시간 3d Reconstruction 이 가능하게 했으며 이를 바탕으로 논문 또한 제출함.
    • 대외비 프로젝트로 전체 코드는 공개 불가.
  • 길건너 친구들 (2018)

    • C++에서 vertex shaderfragment shader 를 사용해 길건너 친구들 게임을 각각 2d 와 3d 로 구현.
  • 그래프 형태의 빅데이터 분석을 위한 알고리즘 연구 및 GPU 프로그래밍 @DGIST (2016)

    • 학부생 3학년 4명이 팀을 이뤄 대학원 연구실에서 제안한 과제를 1년 동안 수행하고 학기 말에 각자의 결과를 발표하는 DGIST 교내 프로그램인 UGRP에서 수행한 과제.
    • 정보통신융합 전공 내 UGRP 프로젝트 1등
    • GPU상에서 아직 구현되지 않은 algorithm 을 구현하고 좀 더 효율적으로 만들 수 있는 방법에 대해 연구한 프로젝트. Betweenness Centrality, Clustering Coefficient, Minimum Spanning Tree and Triangle Counting 의 4개 algorithm 에 대해 연구를 진행했고 이 중 일부 알고리즘들은 parallel computation 에 특화된 GPU 기반으로 코드를 구현함.
    • 맡은 역할 : Develop & Optimize Triangle Counting algorithm
    • UGRP 프로그램 종료 후 본인이 구현한 코드와 알고리즘을 바탕으로 과제를 제안한 연구실의 석박사 과정 대학원생들이 연구를 진행함.
  • 숫자 야구 게임 (2015)

    • Java를 이용해 server와 client 구축 후 2명이 숫자 야구 게임을 즐길 수 있도록 구현.

Fun fact 😏

  • ️반려견을 키우는 집사 🐕 (진주 / 2020.11.13 생일 추정)

  • 오로라 🌠 를 보러 혼자 떠났던 노르웨이

  • 대학교 4년 동안 끊이지 않던 아르바이트 🏃

footer

Popular repositories Loading

  1. 3d-reconstruction 3d-reconstruction Public

    obj loader, synthetic dataset is variable now, but trained CNN is not variable now.

    C++

  2. ShuffleNet-Version2 ShuffleNet-Version2 Public

    Python

  3. Unsharp-masking-Denoising Unsharp-masking-Denoising Public

    Jupyter Notebook

  4. remaining-of-color remaining-of-color Public

    remaining color at 2d image using Matlab

    M

  5. Image-Blending-using-Laplacian-Pyramids Image-Blending-using-Laplacian-Pyramids Public

    Jupyter Notebook 1

  6. Morphological-filter Morphological-filter Public

    MATLAB