Toma Okugawa

About

I’m Toma Okugawa (奥河 董馬), a third-year student in the Information Science and Technology Department at the National Institute of Technology, Yuge College (KOSEN).

I started programming in earnest in spring 2025. An interest in SQL brought me back to Python, and from May I built YUMI-TRACK, a web app that scores kyudo form. It led to a conference presentation and, later, the app Kyudo AI.

My research is low-light image enhancement based on classical Retinex theory. I released FSSR and XCR in 2025, and in 2026 SD-Retinex, which fixes the problems of XCR. The research also powers Beacon, which works with inspection footage from dark places.

Since September 2026 I also make generative art (t-okugawa.studio). I learn in public, failures included: the story so far is in my Medium essay “Seeing in the Dark: How My Research Became Two Startups” and in a KOSEN MONSTER interview (in Japanese). The full list of my work is on researchmap.

Timeline

  1. Paper accepted at IEEE GCCE 2026 (with Keito Inoshita; to appear)
  2. 3rd place, Startup World Cup 2026 Kyushu Regional Youth Pitch Contest (Kyudo AI, team of two)
  3. Fighting Spirit Prize, Original Section, KOSEN Programming Contest 2026 (team of five)
  4. Grand Prize, Lean Launchpad Nagoya 2026 (Kyudo AI, proposed by the three-person team “Beacon”)
  5. Joined the Institute of Electronics, Information and Communication Engineers (IEICE)
  6. NICT Award, JR Central Award, Beyond Next Ventures Award and Tongali Award 5th place, Tongali Business Plan Contest 2026 (Beacon, team of three)
  7. Released the SD-Retinex paper, code and evaluation results
  8. Semifinalist, SusHi Tech Teen Challenge 2026 (a business plan from my sole proprietorship, Second)
  9. Founded Second, a sole proprietorship for programming education by a KOSEN student (closed in May 2026; the lessons became an EdArXiv preprint)
  10. Fighting Spirit Prize, Japanese Olympiad in Informatics (JOI), preliminary round
  11. Overseas programme of LEADING EDGE Shikoku in Singapore; presented FSSR and XCR at NUS
  12. Released the FSSR and XCR preprints on Jxiv
  13. Fighting Spirit Prize, Original Section, KOSEN Programming Contest 2025 (Recipro Engine!, team of five)
  14. Outstanding Presentation Award, 2025 Shikoku-Section Joint Convention of the Institutes of Electrical and Related Engineers (YUMI-TRACK)
  15. Selected as a creator of LEADING EDGE Shikoku, part of METI's AKATSUKI Project (until January 2026), with a 360° vision-assist mobile robot for night-time environments
  16. Joined the Information Processing Society of Japan (IPSJ)
  17. Started building YUMI-TRACK, a web app in Python that scores kyudo form
  18. Outstanding Student Award, Information Science and Technology Department, NIT Yuge College
  19. Passed the Information Technology Passport Examination
  20. Entered the Information Science and Technology Department, National Institute of Technology, Yuge College

Skills and stack

Languages
Python, Dart, TypeScript, SQL
Python for SD-Retinex and YUMI-TRACK, Dart for Kyudo AI, TypeScript for websites and the Kyudo AI server
Vision
NumPy, OpenCV, MediaPipe, ML Kit Pose Detection
Low-light enhancement (Retinex) and pose estimation for kyudo
Apps
Flutter, SQLite (drift)
Kyudo AI (Android)
Web and server
Astro, Streamlit, Cloudflare Workers, D1, Hono
This site and other websites, YUMI-TRACK, the Kyudo AI club server
Tools
Git, GitHub, GitHub Actions
Version control and publishing this site