
Toma Okugawa · Student researcher, NIT Yuge College
Making cameras see in the dark, and phones understand human motion.
I’m Toma Okugawa, a third-year student at the National Institute of Technology, Yuge College, on an island in the Seto Inland Sea. I work on training-free low-light image enhancement (Retinex), real-time pose estimation that runs on the device in front of you, and small language models for regional data, and on turning that research into products through two student ventures, Beacon and Kyudo AI.
Now
A snapshot of what I’m working on, as of September 2026.
- Preparing my talk for IEEE GCCE 2026 in October, on re-defining Japan’s “vanishing municipalities” with multidimensional clustering and small language models, joint work with Keito Inoshita of Kansai University.
- Getting Beacon, our fully offline edge-AI inspection system for dark environments, ready for Kigyoka Koshien, NICT’s national student startup competition, in March 2027.
- Taking Kyudo AI, the smartphone-only kyudo coaching app, from Lean Launchpad Nagoya 2026 toward the national Lean Launchpad final, also in March 2027.
- Continuing the regional-data work: cleaning national statistics, clustering municipalities, and fine-tuning small language models that run entirely on local hardware.
- Studying as a third-year student in the Information Science and Technology Department at Yuge College, on Yuge Island.
Selected research
View allSD-Retinex: A Stable Retinex Method for Low-Light Image Enhancement Designed Around the Sigmoid Derivative
Training-free Retinex enhancement that moves the sigmoid derivative to the real axis: 16.47 dB PSNR on LOL eval15, about 54 FPS at 600×400 on one CPU thread
Ventures & programs
View allBeacon
Closed-network edge-AI inspection for dark factory floors and plant interiors, built on training-free Retinex enhancement; four awards at Tongali BPC 2026
Kyudo AI
Smartphone-only kyudo coaching app that detects the full-draw hold and scores form on the device; Grand Prize at Lean Launchpad Nagoya 2026
NOVA: a 360-degree vision-assisted mobile robot for nighttime environments (LEADING EDGE Shikoku)
Night-vision mobile-robot project selected for LEADING EDGE Shikoku (METI AKATSUKI); it produced the XCR method and a presentation trip to Singapore
Recent recognition
View allLatest updates
View allRe-Defining Vanishing Municipalities in Japan: A Multidimensional Clustering and SLM-Based Policy Insight Framework
Launched this site


