Research and open standards
The Chiba Centre for Visual Science studies how people see, and how devices see on their behalf. It runs the Independent Implant Register, holds the Shared Sight data, and writes the open standards that let any implant be serviced by anyone.
The Centre opened in 2016 in Makuhari, Chiba, a short walk from the bay. It has 61 staff: vision scientists, clinicians, engineers, data stewards and the staff of the Register. It does its own research and funds research elsewhere through the Foundation’s open calls.
Its work falls under three headings: how people with low vision see and how to help them; how the perception software in implants and assistive devices behaves, and how it fails; and how implants should be built so that the people who wear them are not dependent on any one company.
- Head
- Dr Tachibana Shion
- Staff
- 61
- FY2025 spending
- ¥4.87 billion, including research grants
- Research partner
- Frothly, since 2021
- Address
- 2-11 Utase, Mihama-ku, Chiba 261-0013
Publications and datasets
-
Independent Implant Register: annual report for 2025
Reports, signals and serious harm from optical implants of every make in calendar year 2025, with the Register’s estimate of under-reporting and an updated analysis of harm from discontinued devices.
-
Bystanders: what optical implants see of other people
With Oxenfurt University. A survey of what nine implant models record, process and keep about the people their wearers look at, and of the law on it in thirteen countries. The basis for the 2026 revision of KF-OS 2.
-
Lost to follow-up: tracing cataract patients who did not come back
Programme quality team. A pilot in two hospitals in FY2024: of 386 patients traced at home in western Kenya and Nepal after missing follow-up, 68.4% had a good outcome, against 77.6% of those who returned. The method the Foundation adopted across sight restoration in FY2025.
-
Shared Sight: independent audit of anonymisation
An external audit of how Shared Sight removes donors’ identities and blurs other people. It found the method works as described, and made seven recommendations.
-
KF-LV 1.0: a low-vision test set for perception models
With Frothly’s Blackwall programme. 27,500 annotated scenes recorded at eye height in poor light, on stairs, at crossings and in crowds, for testing whether the perception software in assistive devices fails in the places that matter most to blind and partially sighted people.
-
The support gap: discontinued medical implants in six countries
What happens to people when the maker of their implant stops supporting it. Case studies of four device families, with recommendations for makers, regulators and hospitals.
-
Who checks the eye? Independent oversight of optical implants
The final report of the scoping study with Frothly. It recommended independent pre-release evaluation of perception software, firmware escrow for discontinued devices and wider adoption of open service interfaces, and that the Independent Implant Register, which predates the study, be given funding that cannot be cut mid-term. The Foundation has since acted on each.
Publications are free. For a PDF of any of them, or the KF-LV dataset, write to research@kiroshi.org; datasets are released under a data use agreement that forbids re-identifying anyone.
Independent studies of our work
Studies by researchers we did not fund and could not influence, which tested whether our programmes do what we say. We list every one we know of, including those that found us wanting.
-
Displacement in donor-funded cataract surgery: evidence from western Kenya
Compared operations funded by the Foundation in Siaya and Homa Bay counties with public surgical lists before and after the grant. About one in six funded operations would have happened within a year through public provision. The authors found no effect on outcomes.
Ochieng, P. and Varela, T. (2026). Health Systems in Africa, 4(1), 22–40.
-
Would they have got them anyway? A counterfactual study of school spectacle provision in Nepal
Followed children in Mieru and comparison schools in three districts for two years. About a third of children given spectacles by Mieru would have obtained them within two years without it, mostly through private opticians in towns. The effect on children in rural schools was larger than average.
Bhattarai, S., Khadka, R. and Lindqvist, M. (2025). Journal of Global Eye Health, 9(2), 114–131.
-
The Independent Implant Register: an external review of its first seven years
Found that the Register’s signals were timely and its clinical review sound, but that reporting by wearers in the Philippines and Kenya lagged far behind Japan. Recommended aftercare clinics as reporting points, which the Foundation has since adopted.
Commissioned by the board’s Audit and Risk Committee from an independent review team; published in full with the Register’s 2022 annual report.
Open standards
Three standards, free to implement by anyone, written in public with comments from wearers, clinicians and makers.
| Standard | Title | Version | Status | Implemented by |
|---|---|---|---|---|
| KF-OS 1 | Implant service interface | 2.1 | Published 3 July 2023 | 6 of 9 makers |
| KF-OS 2 | Bystander privacy for optical implants | 1.0 | Published 24 September 2024 | 4 of 9 makers |
| KF-OS 3 | Private provenance for ocular capture | Draft 0.4 | Open for comment until 15 January 2027 14 October 2026 | Not yet applicable |
Working with Frothly
Since 2021 the Centre has been a research partner of Frothly, the independent institute that runs the Turing Registry. Together we built KF-LV, the low-vision test set that Blackwall now uses to evaluate perception models in medical implants, and since 2024 Frothly’s Lazarus archive has preserved the firmware of implants their makers no longer support. Frothly’s research on the faithfulness of explanations from vision systems and on provenance for images shaped KF-OS 2 and KF-OS 3.
Mieru at home
The Centre maintains the Landolt chart Mieru uses in schools, and a version you can use at home on a screen or printed. It will not replace an eye test, but it will tell you whether you should have one. Try Mieru at home.