Jinu Lee

Background

About Jinu Lee

I am a Lead AI Research and Machine Learning Engineer. I am researching Efficient Deep Learning, Vision Language Model(VLM), Diffusion, Autonomous Driving, 3D Object Detection, Multimodal, Vision Transformer, AutoML, Neural Architecture Search(NAS), Self-supervised Learning, Robustness Models, and Computer Vision.# Careers 1. Develop Scene Search Platform by VLM and VectorDB (2025.1 ~) 2. Develop Unseen dataset Object Detection and Segmentation by Tags(VLM)(2024.3 ~ 2025.1) 3. Develop Synthetic Generation by VLM for Autonomous Driving (2023.12 ~ 2024.11) 4. Develop 3D Object Detection for Autonomous Driving (2023.8 ~ 2024.9) 6. Develop Efficient Deep Neural Architecture for Face Recognition (2022.1 ~ 2023.3) 7. Develop an Invisible Watermarking System with a Generator from GAN (2021.5 ~ 2021.12) 8. Develop Platform of AutoML Table (2020.9 ~ 2021.8) 9. RL-based Deep Learning to predict bonds with tick history data in Hana Bank (2020.4 ~ 2020.8) 10. Deep Learning to predict funds with high returns in Hana Bank (2018.6 ~ 2020.3)# Skills - Deep Learning for Data Preprocessing & Data Augmentation & Model Evaluation: Python - Deep Learning for Training Models: Pytorch & Tensorflow - Github: https://github.com/jinuvision # Publications 1. Jeonggeun Song, Jinu Lee. Meta Context-Aware Diffusion: Enhancing Generative Modeling through Adaptive Meta Context Integration. On review for CVPR 2025. Hyunseop Kim, Hyo-Jun Lee, Yonguk Lee, Jinu Lee, Hanul Kim, Yeong Jun Koh. GRAE-3DMOT: Geometry Relation-Aware Encoder for Online 3D Multi-Object Tracking. On review for CVPR 2025. Jeong-geun Song, Jinu Lee. ViBid: Linear Vision Transformer with Bidirectional Normalization. UAI 2023. Do-Guk Kim, Jinu Lee. Proxyless Neural Architecture Adaptation at Once. Access 10 (2022). Jongheon Jeong, Sejun Park, Minkyu Kim, Jinu Lee, Doguk Kim, Jinwoo Shin. SmoothMix: Training Confidence-calibrated Smoothed Classifiers for Certified Robustness. NeurIPS 2021. Jinu Lee, Do-Guk Kim, Bohyung Han. Efficient Decoupled Neural Architecture Search by Structure and Operation Sampling. ICASSP 2020- 9 more papers # Patent Pending 1. Jinu Lee, Dongwan Kim, Bohyung Han. Object segmentation method and apparatus for performing the same. Jinu Lee, Jeonggeun Song. Apparatus for data labeling inspection using vision-language model. Jinu Lee, Jeonggeun Song. Method for training end-to-end autonomous driving model using knowledge distillation. Jinu Lee. Apparatus for 3D object detection using virtual lidar data generated by diffusion model- 13 more patents