AI plant ID: how it works and why it's accurate

Our Plant Identifier uses a vision model fine-tuned on thousands of houseplant species to answer the single most asked question in the plant community: what plant is this?

The problem

Every plant group on the internet — Reddit, Facebook, Discord — is half care questions and half "ID help?" posts. People inherit plants from grandmothers, find clippings on the sidewalk, or buy unlabeled nursery stock. Without a name, you can't care for it properly.

How it works

You snap a photo. We send it to our vision model. It returns the most likely species along with care basics: light, water, toxicity, common issues. The whole round-trip takes about 2–3 seconds.

What makes ID accurate

Leaf detail. A clear shot of a single leaf — front and back if possible — beats a wide-angle photo of the whole plant.
Good lighting. Daylight or bright indoor light. Flash bounces and washes out the texture the model relies on.
Plain background. If half the frame is a busy bookshelf, the model spends attention on the bookshelf instead of the plant.
Multiple photos. The identifier accepts several frames at once. More angles, better answer.

What it can't (yet) do

Some genera are notoriously hard. Philodendron and Monstera have dozens of nearly identical juvenile-leaf forms. Variegated cultivars can throw the model off — a variegated Hoya Carnosa "Tricolor" vs "Krimson Princess" comes down to subtle pink-vs-cream variegation that's tough on a phone camera. When the model isn't confident, it tells you so instead of guessing.

Free, no signup

You don't need an account to identify a plant — open the identifier, snap, get an answer. The signed-in experience adds history, the ability to save IDs to a wishlist, and one-tap listing if you want to swap or sell.