Seungkyu Kim

AI Researcher

About Me

Hi, my name’s Seungkyu and I’m currently studying AI for Computer vision.

I am most skilled in: Pytorch and Computer Vision

  • Now I’m looking for a job. This is my portfolio

Projects

Multimodality Leanring for Early Diagnosis of Alzheimer's Disease

Currently Ongoing Project. Will be submitted to Journal.

Alzheimer’s Disease is a severe problem in the world. There is no treatment for stopping or reverse to healthy. Therefore, Early Diagnosis is Crucial. I developed new model for Early Diagnosis of Alzheimer’s Disease using Muli-Modal Image Data.

Water Boundary Segmentation in Satellite Images.

Comparative Study of Deep Learning Model for Semantic Segmentation of Water System in SAR Images of KOMPSAT-5, MJ Kim, SK Kim, DH Lee, JK Gahm, Journal of Korea Multimedia Society, Vol 25., pp. 206-214, 2022

In order to measure the extent of damage from floods and droughts, satellite images are used. For efficiently identify extent of the floods and droughts, We applied various Deep Learning Model for semantic segmentation, and compared models. This Research was submitted to Journal.

Self-Supervised Learning for Alzheimer's Disease

Jan. 2022 - March. 2022

In medical feild, There are relatively few labeled image data. Therefore, Learning without Label is crucial. I used SimCLR, BYOL to Self-Supervised Leanring for Early Diagnosis of Alzheimer’s Disease.

Prediction of Pawpularity (Attractiveness of Pictures)

Oct. 2021 - Dec. 2021

PetFinder is Malaysia’s leading animal welfare platform and they adopt animals. This project aim to make animal happily adopt using predict how much Pictures are attractive. Therefore, I used ViT, ResNet50, EfficientNet and ensemble EfficientNet and ViT.

Education

Pusan National University

MS Computer Science

Mar. 2021 - Present (Expected Graduation Date Feb. 2023)

Pusan National University

BS Mechanical Engineering

Mar. 2013 - Feb. 2019

Experience

LG Electronics

Junior Researcher

Jan. 2020 - Feb. 2021

www.lge.co.kr