Skip to content
View SoongE's full-sized avatar
:octocat:
:octocat:
31 followers · 23 following

Achievements

Achievement: StarstruckAchievement: QuickdrawAchievement: YOLOAchievement: Pull SharkAchievement: Arctic Code Vault Contributor

Achievements

Achievement: StarstruckAchievement: QuickdrawAchievement: YOLOAchievement: Pull SharkAchievement: Arctic Code Vault Contributor

Highlights

Organizations

@Algostu @SWCapstone2021 @gdsc-ajou

Block or report SoongE

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.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

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

Report abuse
SoongE/README.md

Seungmin Oh

Ph.D. Candidate in Artificial Intelligence · Ajou University
Computational Intelligence Lab, advised by Prof. Jongbin Ryu

I research efficient AI across vision, language, and vision–language–action (VLA) models, with a focus on deployment under tight compute budgets from edge devices to robot control loops.

Research Interests

  • Vision–Language–Action models — inference efficiency and amortization for robotic manipulation
  • Efficient LLMs — structured pruning and recovery, low-rank adapter aggregation under quantization
  • Test-time adaptation — training-free adaptation of vision–language models at deployment
  • Efficient architectures — re-parameterization, on-device and low-power computer vision

Selected Work
🏆 1st place, 2025 IEEE Low-Power Computer Vision Challenge, Track 1 (CVPR 2025 Workshop) — on-device image classification deployed on Snapdragon via Qualcomm AI Hub
📄 Published in BMVC, ACCV, Scientific Reports, and International Journal of Surgery
🔬 Ongoing: VLA, structured LLM pruning, test-time adaptation

📚 Experience

Type Date Contents Organization Repository
🧑‍🎓
PhD Program		 2024.03 ~ Now CILab under Prof.Jongbin Ryu Ajou University -
🧑‍🎓
Master’s Program		 2022.03 ~ 2024.02 CILab under Prof.Jongbin Ryu Ajou University -
🥳
Community		 2022.09 ~ 2023.12 GDSC Ajou 1st Core Member Google Developer Student Club GDSC Ajou
🏆 Contest Exhibit 2023.01 ~ Now Google Solution Challenge 2023
(Global Top 100)		 Google Vici-gsc
🏢 Internship 2021.01 ~ 2022.02 CILab under Prof.Jongbin Ryu Ajou University -
🏆 Contest Exhibit 2021.03 ~ 2021.09 Recommender System 기반 검색&아카이브 서비스 "Findmap" 개발
(ACK2021 우수상, 한이음2021 동상)		 한국정보산업연합회,
한이음		 Findmap
🎡 Project 2021.03 ~ 2021.06 NLP를 활용한 스크립트 기반 영상 검색 및 요약 서비스 "FindU" 개발
(DCS2021 대학생 논문경진대회 금상)		 FindU(Ajou University) FindU
🎡 Project 2020.11 ~ 2020.12 레시피 추천 어플리케이션 "Recipe Helper" 개발 TomTomVIP(Ajou Univ) recipe_helper
🎡 Project 2020.09 ~ 2020.10 IsolaA를 이용한 "Autosar" toy project 귀여운 고슴도치(Ajou Univ) toy_autosar
🎡 Project 2020.09 ~ 2020.01 고등학생 전용 어플리케이션 "도담도담" 서비스 출시 DumDum(Ajou Univ) dodam_android
🏢 Internship 2020.07 ~ 2020.08 R&D (Object Detection, OCR, InsightFace) AHAM Corp 🔒 private
🏫 Education 2020.01 ~ 2020.08 Algorithm Study Boradori boradori
🏆 Contest Exhibit 2019.11 ~ 2020.01 금융 문자 분석 경진 대회 KB 금융그룹 NLP/pytorch-models
🏆 Contest Exhibit 2019.08 ~ 2019.12 2019 공개SW 개발자 대회 과학기술정보통신부,
nipa		 chungyo
🏫 Education 2016.03 ~ 2022.02 Department of software Ajou University -

🛠Tech Stack


Pinned Loading

  1. ReCalCon ReCalCon Public

    [BMVC2026] Official source code for Re-calibrated Contrastive Loss for Transformation-Aware Prompt Conditioning in Vision-Language Models.

    1

  2. LPCV2025_Track1_Winning_Solution LPCV2025_Track1_Winning_Solution Public

    [LPCVC2025] Official PyTorch implementation of the 2025 IEEE Low-Power Computer Vision Challenge Track1 Winner at the CVPR 2025 Workshop.

    Python 4

  3. neural_substitution neural_substitution Public

    [ACCV2024] Official source code for Neural Substitution for Branch-level Network Re-parameterization

    Python 2

  4. DBADRP_Classifier DBADRP_Classifier Public

    [IJS2023] Official source code for Deep learning using computed tomography to identify high-risk patients for acute small bowel obstruction: development and validation of a prediction model

    Python 3 1

  5. siamese-one-shot-pytorch siamese-one-shot-pytorch Public

    Example of one shot learning and few shot learning with omniglot dataset.

    Python 27 7

  6. MetaLearning-pytorch MetaLearning-pytorch Public

    Meta learning algorithm using pytorch

    Python 5 3