Waste is visible. The data usually isn’t.

YUK.

Take a photo of waste. YUK helps figure out what it is, whether we’ve seen it before, and what changed after someone acted on it.

A 2026 continuation of the waste-mapping software lineage built for World Cleanup Day.

01

The map should not be a pile of reports. It should be our best current estimate of where waste actually is.

A person, municipality, camera or model can all provide evidence. YUK turns that evidence into observations, links observations that appear to describe the same physical place, and keeps the history instead of overwriting it.

AI is the sensor interpreter, not the authority: identify material, amount, hazard, duplicates and confidence; blur faces and plates; then let people confirm what matters.

02

The project has history

Not a clean slate.
A recovered lineage.

In 2018 the World Cleanup mobile code split during a vendor handoff and an Expo → React Native transition. The old generation survived by accident in a Haamer Ventures fork. In 2026 both surviving Git histories were deliberately reunited before YUK began.

YUK software lineageThe shared World Cleanup mapping project splits into a Haamer Expo and open-source branch and a later World Cleanup React Native branch. Both histories are reunited in 2026 as YUK.World Cleanup mapping2017 → early 2018Haamer / Expo + open source05c2832 · 28 May 2018WCD / React Native1896cb0 · 2 Aug 2018reunited · 2026YUK
2017mapping platform begins
2018two development generations diverge
2026both histories become YUK ancestry
Read the archaeology on GitHub ↗

03

The important rewrite is conceptual

Sites persist.
Observations accumulate.

The old system mostly asked: where are the trashpoints? YUK asks what the evidence currently tells us about waste in the physical world—and what action changed it.

YUK site and observation modelPhotos, people, imports and models create timestamped observations. Observations update a persistent site. Cleanups are interventions that change the site's state without deleting its history.human evidencephoto · report · verificationmachine evidencecamera · model · predictionexternal evidencemunicipality · import · researchOBSERVATIONtimestamped evidence+ confidence + sourceSITEpersistent placecurrent best statecleanup / interventionchanges state, keeps history
01

Camera first

Location is automatic. AI structures the evidence. The person confirms.

02

Offline is normal

Capture should survive ferries, forests, weak networks and old phones.

03

History stays

Cleaning a site changes its state; it does not erase what happened there.

04

Open, with restraint

Public data should be useful without exposing people, sensitive media or unsafe precision.

04

Why now

Waste mapping got cheaper.
Understanding can catch up.

Phones already know where they are and people already photograph strange things in the street. Vision models can now turn a photo into structured evidence cheaply enough to make reporting nearly invisible.

The opportunity is not another cleanup counter. It is an open observational layer that can connect citizen evidence, cleanup outcomes, municipal data and machine sensing into something researchers and communities can actually use.

Early signal

Interested?
Get on the list.

YUK is still at the beginning. Join if you want to test early builds, contribute data or code, research waste, or just see where this goes.