Aims to transform image and video based AI technologies into meaningful services. 💘
- Email : eehoeskrap@gmail.com
- Blog : http://eehoeskrap.tistory.com
- Github : https://github.com/eehoeskrap
- Current Position : AI Research Engineer
- Research Area : 2D/3D Pose Estimation, Object Detection, Human Action Recognition, Image Classification, Avatar & 3D Reconstruction, Object Tracking, Object Segmentation, TensorRT Optimizer, Model Deployment, Deep Learning, Computer Vision 💖💖💖
Work Experience
NAVER Z (Seongnam, South Korea)
- 2022.05 ~ Current 🏃♀️
- AI Research Engineer
KT (Seoul, South Korea)
- 2018.12 ~ 2022.05
- Institute of Convergence Technology
- AI Research Engineer
KETI (Seongnam, South Korea)
- 2018.07 ~ 2018.12
- Human Care System Research Center
- Computer Vision Engineer
Projects
📌 KT
Developed AI technologies using Python and Pytorch, and deployed a model using ONNX and TensorRT.
- Developed AI traffic solution for C-ITS including robust object detectors, a Pedestrian Detection System, a Vehicle Detection System, etc. (2020 ~ Current)
- Developed “helmet-wearing detection” issues by researching and developing a human detector, a pose estimator, and a helmet classifier. (2021 ~ Current)
- Researched and developed human detecting, pose estimating, and action/object classifying technologies to develop “Action Recognition” solutions for businesses. (2019 ~ 2020)
📌 Personal Projects
Started a personal project to develop skills with interest in 3D Pose Estimation and Action Recognition.
- The use of 2D Pose Estimation for action recognition has many difficulties due to the Occlusion Problem. Therefore, 3D Pose Estimation was attempted. In order to solve the jitter, the noise was minimized by filtering the pose. (2020.12)
- Developed a counting system by classifying push up/down motions based on 2D & 3D pose estimation. (2021.05)
📌 KETI
Developed computer vision algorithms using C++ in an Linux environment.
- Developed a “Skin Lesion Recognition” algorithm by developing hair removal and mole detection functions. (2018)
- Developed a capture program using FLIR’s Spinnaker camera firmware library. (2018)
📌 Kyonggi University, Graphics & Image Processing Lab
Developed several computer vision algorithms using the OpenCV library and C++.
- A binocular image-based, moving object detection and tracking system for a CCTV environment. (2016 ~ 2018)
- A machine vision based LED panel defect detection system. (2017 ~ 2018)
- An automatic pest-counting system, based on image analysis with Cesco Co., Ltd. (2016)
Skills
- Language : Python, C/C++
- Library : Pytorch, TensorRT, OpenCV, DeepStream, TensorFlow, TensorFlow Lite
- Etc. : ONNX, CUDA, …
Education
Kyonggi University (2017.03 ~ 2018.08)
- Master of Computer Science
- Computer Graphics & Image Processing Laboratory (2017.03 ~ 2018.06)
- Thesis : RGB-D 정보를 이용한 객체 탐지 기반 3차원 인간 자세 추정
- GPA : 4.50/4.50
Kyonggi University (2013.03 ~ 2017.02)
- Bachelor of Computer Science
- Computer Graphics & Image Processing Laboratory (2014.06 ~ 2017.02)
Maehyang Girls’ Information High School (2010.03 ~ 2013.02)
- Department of Digital Contents
- The first time I learned Image Processing through Visual Basic programming was a key event that I decided to become a Computer Vision Researcher.
Publications
Journals
- Seohee Park, Myunggeun Ji and Junchul Chun, “A Method for 3D Human Pose Estimation based on 2D Keypoint Detection using RGB-D information”, Journal of Internet Computing and Services(JICS), 2018. 12. KCI
- Seohee Park, Myunggeun Ji and Junchul Chun, “2D Human Pose Estimation based on Object Detection using RGB-D information”, KSII Transactions on Internet and Information Systems(TIIS), 2018.02. SCIE
- Seohee Park and Junchul Chun, “인간 행위 인식을 위한 비전 기반 인간 자세 추정에 관한 연구”, 인터넷정보학회지, Vol. 18, No. 2, pp. 19-25, 2017.12.
- Seohee Park and Junchul Chun, “A Method for Body Keypoint Localization based on Object Detection using the RGB-D information”, Journal of Internet Computing and Services(JICS), 2017. 12. KCI
- Seohee Park and Junchul Chun, “A Robust Object Detection and Tracking Method using RGB-D Model”, Journal of Internet Computing and Services (JICS), 2017.08. KCI
- Seohee Park, and Junchul Chun, “CCTV 영상의 동적 객체 탐지 및 추적 기술 동향”, 인터넷정보학회지, Vol. 18, No. 1, pp. 39-43, 2017.06.
Conferences
- Seohee Park and Junchul Chun, “3D Human Pose Estimation from 2D Joint Detection via Deep Learning”, Proceedings of 13th APIC-IST 2018, pp. 305-306, 2018.06.
- Seohee Park, Myunggeun Ji, etal., “머신 비전 기반 LED 판넬 결함 자동 검출 시스템”, 한국인터넷정보학회 춘계학술발표대회 논문집 (제19권 1호), 2018.04.
- Seohee Park, Myunggeun Ji, Junchul Chun, “컨볼루션 신경망을 통한 2차원 키포인트 탐지 기반 3차원 인간 자세 추정 방법”, 한국인터넷정보학회 춘계학술발표대회 논문집 (제19권 1호), 2018.04. 최우수논문상 🎉
- Seohee Park, Myunggeun Ji, Junchul Chun, and Byoungdai Lee, “A Defect Detection Method of LED panel based on Image Analysis”, Proceedings of the 9th International Conference on Internet (ICONI 2017), pp. 129-130, 2017.12.18.
- Seohee Park, Myunggeun Ji, etal., “영상분석기반 LED 패널 결함 검출 시스템”, 한국인터넷정보학회 추계학술발표대회 논문집 (제18권 2호), pp. 141-142, 2017.11.
- Seohee Park and Junchul Chun, “3D CCTV based Object Detection and Tracking using RGB-D information”, Proceedings of 12th APIC-IST 2017, pp. 274-276, 2017.06. Outstanding Paper Award 🎉
- Seohee Park and Junchul Chun, “3차원 CCTV 기반 이동 객체의 자동 탐지 및 추적에 관한 연구”, 한국인터넷정보학회 춘계학술발표대회 논문집 (제18권 1호), pp. 123-124, 2017.04. 우수논문상 🎉
- Seohee Park and Junchul Chun, “3차원 CCTV 기반 이동 객체의 자동 추적 방법”, 한국인터넷정보학회 추계학술발표대회 논문집 (17권 2호), pp. 295-296, 2016.11. 학부 최우수논문상 🎉