Chang Wook Seo

I am a PhD candidates at KAIST (Korea Advanced Institute of Science and Technology), Culuture Technology advised by Junyong Noh.

Master of Science at Sogang University, Media Technology, advised by Yongduek Seo. Bachelor of Fine Arts at Kyoto University of Arts.

Before starting my research career as a computer scientist from master's course, I was a 3D artist at Donuts Co. Ltd. I've worked on game develop team Warship Craft.

Email  /  CV  /  Google Scholar  / 

profile photo
Research

I'm interested in computer vision, graphics and deep learning. Most of my research is about investigating better applications to creative domains.

Semi-supervised reference-based sketch extraction using a contrastive learning framework
Chang Wook Seo, Amirsaman Ashtari, Junyong Noh
SIGGRAPH, 2023, ACM Transactions on Graphics (TOG)
project page / Paper

We proposed a reference-based sketch extraction model which can be trained with unpaired dataset via contrastive learning framework.

Reference Based Sketch Extraction via Attention Mechanism
Amirsaman Ashtari*, Chang Wook Seo*, Cholmin Kang, Sihun Cha, Junyong Noh
SIGGRAPH ASIA, 2022, ACM Transactions on Graphics (TOG)
* denotes contributed equally to this work and are listed alphabetically
project page / Paper

We proposed a reference-based sketch extraction model which imitates the style of reference image to output sketch.

Seg2pix: Few Shot Training Line Art Colorization with Segmented Image Data
Chang Wook Seo, Yongduek Seo
Applied Sciences, 2021
Paper

We proposed line art auto-colorization model which enhances its performance by using a few segmented image data.




Project

In project, I've developed many different kinds of applications based on deep learning. Mesh reconstruction, image recognition, image manipulation, video generation, video captioning and more.
Moreover, I also have the experience of creating 3D resources for mobile game as a 3D artist.

KOCCA: 3D Fashion avatar reconstruction for metaverse
June 2023 ~ Present
I'm developing the AI model for 3d human reconstruction.
Additionally researching on fashion image manipulation techniques.

NIA: Construct dataset for video captioning
June 2023 ~ Present
I'm developing the AI model for video captioning.

Samsung Research: Future project
June 2022 ~ Dec 2022
I developed new video generation techniques for future project.

IITP: Cinemagraph AR/VR project
Mar 2021 ~ Dec 2022
I developed scene mesh reconstruction method for VR devices.
Additionally researched 3D human generation based on a single image.

Hyundai Mnsoft: SD map road status generation
Mar 2019 ~ Dec 2019
I developed image classification method for traffic sign plate
Additionally researched OCR (Optical Character Recognition) method for recognizing traffic sign plate (Korean)

Game: 艦つく -Warship Craft-
Apr 2017 ~ Jan 2019
I created 3D modeling and animation resources of the mobile game.
Additionally developed in-house tool of MAYA and Unity shader.