How PlantiD works, in plain terms.

Every result comes with a confidence score. Above 90% usually means the model is confident in the species; anything below 70% should be treated as a starting point and cross-checked. Clear, well-lit photos of the whole plant (leaf, stem, and any flowers) give the best results.

Yes — it flags disease, pest damage, nutrient deficiency, and environmental stress, rates severity from Low to Severe, and suggests treatment and prevention steps. For high-value crops or ambiguous symptoms, confirm with a local extension officer or plant pathologist before acting.

Drone Mode accepts a batch of aerial or walk-through photos of a plot and returns a plot-level health read on the dominant crop, rather than a single-specimen identification. Useful for farm surveys where you care about coverage and stress patterns instead of one leaf.

Some species (e.g. Cannabis, Opium Poppy) are regulated in most jurisdictions. When the model is confident a photo shows one, we surface it explicitly so regulators, land managers, and growers all have the same signal to act on.

Yes — photos and analysis results are stored to your private History so you can return to them, export them, and give feedback. You can delete any analysis (and its photos) from the History page at any time.

Yes. From an analysis result or a row in History, use “Export report” to download a PDF you can share with a client, extension officer, or your own records.

The 👍 / 👎 on a result is stored against the analysis and used to prioritise where the identification and health-assessment prompts need work. It is never shared publicly.

During the beta, PlantiD is free to use with a fair-use limit on analyses per day so a single account can’t swamp the vision model. If you hit the limit, come back in an hour.