Moon Ye-Bin

I am a Ph.D. student in Algorithmic Machine Intelligent (AMI) Lab, Dept. of Electronic Engineering, POSTECH, South Korea, advised by Prof. Tae-Hyun Oh. In the first half of 2025, I will start a research internship in London, U.K., at Huawei Noah's ArkLab - Multimodal team. I finished my master's degree from AMILab in POSTECH, and bachelor's degree from Dept. Electrical and Electronic Engineering at Chung-Ang University (CAU), South Korea.

I am interested in data-centric model training and evaluation, especially within vision and language multi-modal characteristics.
Keywords: multi-modal learning, large vision-language models (LVLMs), large language models (LLMs), but not limited to

Contact: ybmoon[at]postech[dot]ac[dot]kr | aileenmoon96[at]gmail[dot]com
Information: CV, GitHub, Google Scholar

Publications

BEAF: Observing BEfore-AFter Changes to Evaluate Hallucination in Vision-language Models, ECCV'24
Moon Ye-Bin, Nam Hyeon-Woo, Wonseok Choi, Tae-Hyun Oh

[Project Page][Dataset][GitHub][arXiv]

The key idea of our BEAF benchmark is manipulating visual scene information and designing the metrics based on the model's answer changes along the scene changes.

VLM’s Eye Examination: Instruct and Inspect Visual Competency of Vision Language Model, Submitted
Nam Hyeon-Woo, Moon Ye-Bin, Wonseok Choi, Lee Hyun, Tae-Hyun Oh
[arXiv]

We propose an eye examination process to investigate how a VLM perceives images, specifically focusing on key elements of visual recognition, from primitive color and shape to semantic levels.

SYNAuG: Exploiting Synthetic Data for Data Imbalance Problems, ICCV '23 Workshop
Moon Ye-Bin, Nam Hyeon-Woo, Wonseok Choi, Nayeong Kim, Suha Kwak, Tae-Hyun Oh
[arXiv]

Propose to utilize synthetic data to address three distinctive data imbalance problems
[Workshop] ICCVw 2023 (MMFM: What is Next in Multimodal Foundation Models?)

TextManiA: Enriching Visual Feature by Text-driven Manifold Augmentation, ICCV'23
Moon Ye-Bin, Jisoo Kim, Hongyeob Kim, Kilho Son, Tae-Hyun Oh
[Project Page][GitHub][Paper][arXiv]

Propose a text-driven manifold augmentation method that semantically enriches visual feature spaces, regardless of data distribution
[Workshop] CVPRw 2023 (WFM: Workshop on Foundation Models)
[Workshop] ICCVw 2023 (MMFM: What is Next in Multimodal Foundation Models?)

ENInst: Enhancing Weakly-supervised Low-shot Instance Segmentation, Pattern Recognition (IF=8.0)
Moon Ye-Bin, Dongmin Choi, Yongjin Kwon, Junsik Kim, Tae-Hyun Oh
[Paper][arXiv]

Proposed ENInst with sub-task enhancement methods: instance-wise refinement for enhancing pixel localization quality and novel classifier composition for improving classification accuracy, based on the performance bottleneck analyses
[Conference] IPIU 2022, Best Paper Award

HDR-Plenoxels: Self-Calibrating High Dynamic Range Radiance Fields, ECCV'22
Kim Jun-Seong, Kim Yu-Ji, Moon Ye-Bin, Tae-Hyun Oh
[Project Page][Paper][arXiv]

Propose High Dynamic Range Plenoxels (HDR-Plenoxels) that learns the plenoptic function of the 3D scene from a comprehensive understanding of 3D information, physical radiance field, and varying camera settings inherent in 2D low dynamic range (LDR) images
[Workshop] CVPRw 2023 (3DMV: Learning 3D with Multi-View Supervision)

FedPara: Low-rank Hadamard Product Parameterization for Efficient Federated Learning, ICLR'22
Nam Hyeon-Woo, Moon Ye-Bin, Tae-Hyun Oh
[Paper][arXiv]

Proposed a communication-efficient re-parameterization in the federated learning setting

PROJECTS

Localized Image Retrieval, Samsung Research, 2024
Time Series Regression with Large Language Models, SAIT, 2024
Abnormal and Danger Signs Detection with LLMs, KRIT, 2023
Video Panoptic Segmentation and Depth Estimation, ETRI, 2022
Data Augmentation for Domain Adaptive Object Detection, LG Display, 2022
Weakly-supervised Low-shot Instance Segmentation, ETRI, 2021
Self-supervised Few-shot Learning by Episodic Instance Discrimination, ETRI, 2020

EDUCATION

Ph.D. in Electrical Engineering, POSTECH, Pohang, South Korea (current)
- Mar.2022 -
Master in Electrical Engineering, POSTECH, Pohang, South Korea
- Sports AIX Program
- Mar. 2020 –Feb. 2022
Bachelor in Electrical and Electronic Engineering, Chung-Ang University, Seoul, South Korea
- Magna cum laude
- Mar. 2016 – Feb. 2020

EXTRA

Conference Reviewer
- WACV'25
- ACCV'24, ECCV'24, CVPR'24
- ICCV'23, CVPR'23, WACV'23
Teaching Experiences
- [NAVER Boostcamp AI Tech 6th] Data-centric CV course, NAVER & Upstage, 2023
- [EECE236] Learning Electronic Engineering with MATLAB, POSTECH, 2020
Deep Learning Seminar
- POSTECH AMILab seminar [YouTube]
- POSTECH EE-group deep learning seminar
- POSTECH PAIR (CVLab, CGLab, MLLab, AMILab) seminar

Moon Ye-Bin

BEAF: Observing BEfore-AFter Changes to Evaluate Hallucination in Vision-language Models, ECCV'24Moon Ye-Bin*, Nam Hyeon-Woo*, Wonseok Choi, Tae-Hyun Oh

[Project Page][Dataset][GitHub][arXiv]

VLM’s Eye Examination: Instruct and Inspect Visual Competency of Vision Language Model, SubmittedNam Hyeon-Woo, Moon Ye-Bin, Wonseok Choi, Lee Hyun, Tae-Hyun Oh[arXiv]

SYNAuG: Exploiting Synthetic Data for Data Imbalance Problems, ICCV '23 WorkshopMoon Ye-Bin*, Nam Hyeon-Woo*, Wonseok Choi, Nayeong Kim, Suha Kwak, Tae-Hyun Oh[arXiv]

TextManiA: Enriching Visual Feature by Text-driven Manifold Augmentation, ICCV'23Moon Ye-Bin, Jisoo Kim, Hongyeob Kim, Kilho Son, Tae-Hyun Oh[Project Page][GitHub][Paper][arXiv]

ENInst: Enhancing Weakly-supervised Low-shot Instance Segmentation, Pattern Recognition (IF=8.0)Moon Ye-Bin, Dongmin Choi, Yongjin Kwon, Junsik Kim, Tae-Hyun Oh [Paper][arXiv]

HDR-Plenoxels: Self-Calibrating High Dynamic Range Radiance Fields, ECCV'22Kim Jun-Seong*, Kim Yu-Ji*, Moon Ye-Bin, Tae-Hyun Oh [Project Page][Paper][arXiv]

FedPara: Low-rank Hadamard Product Parameterization for Efficient Federated Learning, ICLR'22Nam Hyeon-Woo, Moon Ye-Bin, Tae-Hyun Oh [Paper][arXiv]