Hi, there.

My name is Jinseo Jeong.

I study image formation for controllable visual generation,
with a focus on light transport.

About me

I’m a Ph.D. student in Computer Science and Engineering at Seoul National University, advised by Professor Gunhee Kim in the Vision & Learning Lab.

My research connects scene conditions to visual appearance through image formation and light transport. Building on my work in inverse rendering, neural rendering, and diffusion-based object compositing, I’m exploring how generative models can realize intended visual effects together with the scene conditions that produce them.

Earlier in my Ph.D., I worked on continual learning with noisy and imbalanced data streams. I did my undergraduate degree at Chung-Ang University, graduating as valedictorian of the College of Engineering.

Download CV (PDF)
Jinseo Jeong

Publications

NeurIPS 2026

LOCO: Local Light-Aware Object Compositing with Spatially Varying Illumination-Augmented Data

Jinseo Jeong, Hyunsoo Kim, Junseo Koo, Junhyeog Yun, Gunhee Kim, and Soo Ye Kim
Joint work with Adobe Research

Makes an inserted object match the local lighting of a photo, using training data augmented with spatially varying illumination.

LOCO: Local Light-Aware Object Compositing with Spatially Varying Illumination-Augmented Data
AAAI 2026

Gaussian Blending: Rethinking Alpha Blending in 3D Gaussian Splatting

Junseo Koo, Jinseo Jeong, and Gunhee Kim

Replaces scalar alpha blending in 3D Gaussian Splatting with spatially varying alpha and transmittance, so fine detail survives zooming in and out.

Gaussian Blending: Rethinking Alpha Blending in 3D Gaussian Splatting
CVPR 2024

ESR-NeRF: Emissive Source Reconstruction Using LDR Multi-view Images

Jinseo Jeong, Junseo Koo, Qimeng Zhang, and Gunhee Kim

Reconstructs in-scene emissive sources from LDR multi-view images, separating emitted light from reflected light.

ESR-NeRF: Emissive Source Reconstruction Using LDR Multi-view Images
ICCV 2021

Continual Learning on Noisy Data Streams via Self-Purified Replay

Jinseo Jeong*, Chris Dongjoo Kim*, Sangwoo Moon, and Gunhee Kim (*equal contribution)

Keeps replay-based continual learning robust to noisy labels: self-supervised learning limits forgetting, and a purified replay buffer keeps label noise out.

Continual Learning on Noisy Data Streams via Self-Purified Replay
ECCV 2020

Imbalanced Continual Learning with Partitioning Reservoir Sampling

Chris Dongjoo Kim, Jinseo Jeong, and Gunhee Kim

Extends continual learning to multi-label, long-tailed data streams with a replay sampling strategy that keeps both head and tail classes represented.

Imbalanced Continual Learning with Partitioning Reservoir Sampling

Experience

Industry Collaboration
Adobe Research
2025 – Present

Diffusion-based object compositing under spatially varying illumination; first-author paper LOCO accepted to NeurIPS 2026.

Industry Collaboration
Samsung Electronics MX (Mobile Experience)
2022 – 2024

Inverse rendering for virtual object insertion (2023–2024); image synthesis under novel viewpoints and lighting (2022–2023).

Research Intern
Vision & Learning Lab
Jan 2019 – Feb 2020
Seoul National University, Seoul, Korea

Advised by Professor Gunhee Kim. Worked on classifying imbalanced display-panel defect datasets; the resulting paper (IMID 2020) won the Samsung Display Paper Award.

Research Intern
Intelligent Information Processing Lab
Apr 2018 – Aug 2018
Chung-Ang University, Seoul, Korea

Advised by Professor Youngbin Kim. Published Content Preserving Style Transfer via Foreground Separation Algorithm (KCGS 2018).

Research Intern
Motion Graphics Lab
Dec 2016 – Feb 2018
Chung-Ang University, Seoul, Korea

Advised by Professor Kyoungju Park. Published 360 Image based Mixed Reality and User Interaction (HCI Korea 2018).

Education

Integrated M.S. / Ph.D. in Computer Science and Engineering
Seoul National University
Mar 2020 – Present
Seoul, Korea

Advised by Professor Gunhee Kim, Vision & Learning Lab.

B.S. in Integrative Engineering
Chung-Ang University
Mar 2015 – Feb 2020
Seoul, Korea

Valedictorian, College of Engineering (GPA 4.43 / 4.5). Graduated one semester early (7 semesters).

Awards and Honors

Industry
Samsung MX Appreciation Award for Industry-Academia Project, 2023

Industry-academia project on virtual object insertion, with Junseo Koo and Gunhee Kim.

Industry
Samsung Display Paper Award (Gold Prize), 2020

For Deep Knowledge Transfer for Classification of Imbalanced Defect Image Datasets of Display Panels (IMID 2020), with Minui Hong, Jinsoo Yu, and Gunhee Kim.

Volunteer
Deputy Prime Minister and Minister of Education Award, 2019

Outstanding education volunteering; team leader.

Professional Activities

Reviewer
  • Conferences: CVPR 2024–2026, CVPRW 2021, ECCV 2026, NeurIPS 2025–2026, AAAI 2025, BMVC 2026, ACM MM 2025–2026 (incl. Dataset Track 2026)
  • Journal: IEEE TCSVT 2025
Invited Talk

Bringing Artificial Intelligence to Mixed Reality, Department of Artificial Intelligence, Chung-Ang University, 2024.

Contact

For research collaborations or internship opportunities, email is the best way to reach me.

Vision & Learning Lab, Seoul National University · Seoul, Korea