Parki ar Minda · Digital tool

What I Recycle

Photograph an item and get a suggested category under Eco-taxi rules—before you bag it for pickup or a garage drop-off.

Built around the same fractions we publish in our guidelines.
Use the web scanner in a few taps, or the mobile app if you prefer a native flow.
Helpful when you are new to sorting or when a package is easy to misread.

Open web scanner

Keep the app on your phone

What I Recycle app: camera capture and category result on two phone screens

Why You’ll Love It

Instant clarity — stop wondering whether that yogurt cup is recyclable
Beginner-friendly — perfect if you’re just starting to sort your waste
Always up to date — sorting guidelines are updated as Eco-taxi expands its accepted materials
Works with Eco-taxi — sort at home, then schedule a pickup in one tap

At a glance

Crushed plastic water bottle held in hand ready for recycling

Photo in

Camera or gallery on web and mobile.

Plastic PET container held in hand, app recognition box labelled 01 PET other

Category out

Matches our published sorting streams.

Same Eco-taxi world

Guidance lines up with what we collect at garages and on pickup routes.

A Note on Privacy

To return a sorting suggestion, your photo is sent to Google’s cloud AI (Gemini) for that request. We do not keep a copy on our own systems: once the result is returned, the image is discarded from our infrastructure. We do not use your photos for training, marketing, or any secondary processing—only account data needed for sign-in and scan limits is stored.

Full detail: Privacy policy · Terms & conditions

Web allowance and paid scans

The website offers a limited number of free scans per month (with a Google sign-in). If you need more, scan packs are purchased in the mobile app. You can then link that device to your web account and use the same balance in the browser.

Why limits at all: each cloud classification has a direct cost. Caps and paid packs keep the service within budget.

Background

In 2023 we launched “What I Recycle” with an ambitious idea: build a local, offline waste-recognition model trained on pictures contributed by our community. The concept was sound, but collecting enough labeled data proved too difficult — user participation simply wasn’t sufficient to train a reliable model.

Fortunately, AI caught up. Large Language Models (LLMs) now handle visual classification with remarkable accuracy — no custom dataset required. So we rebuilt the app from the ground up.