I plan to continue uploading the projects I have done in the future.
I’m currently learning multi-modal, multi-camera, lidar.
I’m looking to collaborate on everything about computer vision.
Ask me about everything about me.
How to reach me byul3325@naver.com or byul3325@gmail.com.
my github blog git blog.
I have experience collaborating with companies here !!! About 2D/3D Object Detection, 3D Reconstruction(Bundle Adjustment, Pose Graph Optimization), On-board(memory & time optimization), Camera Calibration(using 1D/2D chessboard), AI ethics, AI module performance improvement in specific environment ...
Brief description of the projects involved 📚
- Goal: Develop a 3D reconstruction module using monocular images.
- Role:Lead Researcher (80% contribution) – Designed and implemented keypoint matching (SIFT/SURF), computed epipolar geometry, estimated camera relationships, and applied PnP & bundle adjustment (BA). Led the development of the full 3D reconstruction pipeline.
- Achievement: Successfully built a 3D reconstruction module that processes monocular images to generate 3D structures.
2. Robust Monocular Camera 3D Object Detection in Various Camera Environments (Hyundai) 🎯 (Mar 2021 - Jun 2022)
- Goal: Improve the robustness of monocular camera-based 3D object detection, addressing significant performance degradation caused by varying camera environments.
- Role:Lead Researcher (70% contribution) – Developed data augmentation techniques to enhance model generalization across different camera angles, identified the root causes of performance degradation, and implemented correction algorithms to mitigate these effects.
- Achievement: Diagnosed key factors affecting model accuracy and significantly improved performance:
- Accuracy increased from 20% to 80% for a 3-degree angle variation.
- Accuracy increased from 1% to 50% for a 5-degree angle variation.
- Research findings contributed to international patents and publications(CVPRw 2024).
3. Development of Car Location and Speed Estimation Module Using CCTV Footage (ETRI) 🚗📹 (Aug 2022 - Dec 2022)
- Goal: Develop a module capable of estimating vehicle position and speed solely from CCTV video data.
- Role:Lead Researcher (80% contribution) – Engineered road detection and image warping algorithms, developed vehicle speed estimation methods, and optimized overall system performance.
- Achievement: Achieved over 90% accuracy in vehicle speed estimation on the target dataset.
- Miltitary Scientific Surveillance System(육군본부) - 23.03 ~ 23.09
- Performance Enhancement Officer (30% contribution)
- 목표 : AI 경계감시시스템 구축을 통해 오탐지/미탐지 감소 및 정탐지 향상
- 역할 : TOD 카메라 정보(카메라로부터 절대 거리 가능)를 활용하여 오탐지를 획기적으로 줄였음
- 성과 : 기존 대비 오탐지를 10% 줄였음
- AI 무기체계 시험평가 기준 설립(육군본부, 미국방부) - 23.03 ~ 24.06
- AI 시험평가 연구원 (30% contribution)
- Goal : Develop new testing and evaluation standards for AI weapon systems, which differ significantly from traditional weapon systems.
- Role : As an AI Test and Evaluation Researcher, collaborated with the U.S. Department of Defense, coordinated with the Ministry of National Defense, and conducted extensive research on AI weapon systems, including identifying requirements (Contribution 30%).
- Achievements : Established initial standards for the Military Performance Certification Center (including dataset construction, baseline model development, and formulation of various strategies).
- AI 시험평가 기준 모델 연구개발(육군본부) - 23.10 ~ 24.06
- Lead Researcher (60% contribution)
- Goal : When introducing various AI weapon systems in the Army, create a military learning/test data set and build an AI model that serves as a standard.
- Role : Developed an Auto-Labeler for military datasets, performed data cleansing and construction, and developed a baseline model for performance verification (Contribution 60%).
- Achievements : Established initial standards for the Military Performance Certification Center(including dataset construction, baseline model development, and formulation of various strategies).
- 프로젝트 이름 찾기(휴비츠) - 24.06 ~ 24.12
- 핵심 개발자 (40% contribution)
- 목표 :
- 역할 : Multi-Thread와 자료구조 개선을 통한 최적화
- 성과 : 성능 저하 없이 기존 알고리즘 대비 최대 60% 속도 향상
- 프로젝트 이름 찾기(현대조선해양) - 24.09 ~ 24.11
- 핵심 개발자 (40% contribution)
- 목표 :
- 역할 : 원형격자 chess board를 통한 camera calibration 모듈 개발
- 성과 : ~(수치 찾아두기)의 정확도로 원형 chess board 활용하는 camera calibration 모듈개발
- 알약 검출 및 인식(ETRI) - 24.09 ~ 24.12 => 문서 찾아보기
- Lead Researcher (80% contribution)
- 목표 :
- 역할 : 추가적인 2D 알약 검출 알고리즘 학습없이 알약을 검출하고 어떠한 알약인지 인식하는 알고리즘 개발(Template matching, color 고려, warping 등)
- 성과 :
- 📫 You can edit your github read me in https://rahuldkjain.github.io/gh-profile-readme-generator/