Kiyoon Kim

AI Platform Lead at Deargen
PhD in Computer Vision AI at The University of Edinburgh

Find Out More

I've got skills!

I started computer programming when I was 9. I began MFC(C++) Windows programming at age 14 and opened a personal website, "sparkware.co.kr". I made a Hair Proportion Analysis algorithm and obtained a patent when I was 17. Now my interests are in machine learning application on computer vision and image signal processing.

"Kiyoon" stands for glowing field. I want to make my field of study glow in the future.

Cirriculum Vitae

History of My Life

  • 2004 (9 years old)

    First Programming

    I started computer programming with Flash ActionScript 2.0.

  • 2010 (15 years old)

    MFC Windows Programming

    I started MFC Windows Programming and opened my own website, sparkware.co.kr. I released many Windows programs on this website.

  • 2012 (17 years old)

    Patent : Hairloss Analysis Algorithm

    I finished a project about Hair Proportion Analysis algorithm based on the Maximum Likelihood estimation.

  • 2014-2018 (18-22 years old)

    Electrical Engineering & CS @ UNIST, won many contests, graduated with Chairman's Award!

    I studied both Electronics and Computer Engineering. I won lots of competitions like Hexathon (supported by Naver), Korea Supercomputing Challenge, and Naver Poster Award. See experience below.

  • 2016-2018 (20-22 years old)

    Started working as a researcher at UNIST, EgoVid Inc.

    We've visited UC Berkeley for 9 weeks to exchange ideas about making a start-up company. (Article)
    Extreme Low Resolution Activity Recognition Using Siamese Embedding (Paper)
    We've demonstrated our technologies at CVPR 2017 Expo. (Youtube video)

  • 2018-2023 (22-28 years old)

    MSc in Artificial Intelligence at The University of Edinburgh

    PhD in Informatics at The University of Edinburgh

  • There will
    be more
    story!

Publications

K Kim, SN Gowda, P Eustratiadis, A Antoniou, RB Fisher, “Adversarial Augmentation Training Makes Action Recognition Models More Robust to Realistic Video Distribution Shifts”, 4th International Conference on Pattern Recognition and Artificial Intelligence, ICPRAI 2024

K Kim, D Moltisanti, O Mac Aodha, L Sevilla-Lara, “An Action Is Worth Multiple Words: Handling Ambiguity in Action Recognition”, 33rd British Machine Vision Conference 2022, BMVC 2022

K Kim, SN Gowda, O Mac Aodha, L Sevilla-Lara, “Capturing Temporal Information in a Single Frame: Channel Sampling Strategies for Action Recognition”, 33rd British Machine Vision Conference 2022, BMVC 2022

SN Gowda, L Sevilla-Lara, K Kim, F Keller, M Rohrbach, “A new split for evaluating true zero-shot action recognition”, DAGM German Conference on Pattern Recognition, 2021

M. S. Ryoo, K. Kim and H. J. Yang, ”Extreme Low Resolution Activity Recognition with Multi-Siamese Embedding Learning, ”AAAI Conference on Artificial Intelligence, New Orleans, Louisiana, February 2018. [acceptance rate: 24.6%]

Awards

1st place - Naver UNIST Undergraduate Poster Award
1st place - HeXATHON, UNIST, supported by NAVER
KSC (Korea Supercomputing Challenge) : MPI parallel computing 5th place
Bronze Medal in the National Olympiad for Informatics (KOI, Korea Olympiad in Informatics)
Patent : Hair Percentage Analysis algorithm
UNIST Startup clinic: Smart home app controlling electrical output

Languages, Frameworks and Tools:

Languages: Python, TypeScript, C, C++, C#, Lua, Bash
Deep Learning: PyTorch, Keras
Parallel Programming: MPI, CUDA, PyTorch DDP (Distributed Data Parallel)
Tools: Git, GitHub, Docker, Kubernetes
Research: MATLAB, OpenCV
Embedded Computer: NVIDIA Jetson TX2, Raspberry Pi (IoT)
Database & Web: React, PHP, MySQL, PostgreSQL, FastAPI, Flash ActionScript, WordPress
Others: NXT Robot C, Electron/WPF/MFC (desktop app)

Let's Get In Touch!

kiyoonkim@[remove-this]deargen.me