Scout every leaf.
Catch disease early.

The robot's eyes, in your browser. The camera is read a few times a second; each frame goes through the crop check, the condition model and four safeguards, and a result shows only when several frames agree.

Checks tomato and cucumber, leaf or fruit: first which crop and part it is, then the common diseases and problems of that part.

01 / Input

Camera · photo

Show it one leaf

Use the live camera, or one photo of a single leaf or fruit in daylight.

Experimental models. Confirm the crop and part before acting. Whole plants, seedlings and rare diseases may be missed or not covered. This is not a diagnosis.

No photo yet

Open the live camera, or drop one clear photo here

Analysed on this device · never uploaded · JPEG, PNG or WebP up to 15 MB

For condition screening: one close, sharp leaf or fruit filling the square.

02 / Result

On this device

Waiting for a photo

The result will appear here

Show it a tomato or cucumber leaf or fruit. The models run in this browser; the photo isn't sent anywhere.

1Frame one leaffill the square · daylight · sharp focus
2Hold steady3 of the last 4 frames must agree
3Read the resultcondition · confidence · spray estimate

Screening support, not a laboratory diagnosis. The spray estimate uses demo rates, not a product's label rates: confirm uncertain or severe symptoms with an agronomist and read the label before treating.

Model release: CropScout Vision v31. Our own models recognise the crop and check tomato leaves, tomato fruit and cucumber fruit, trained with photos of leaves on the plant from four public field datasets; cucumber leaves still use the older v26 model. The live camera and the spray estimate come from CropSCOUT Live, built by our team at HackHarvard. Source code (AGPL-3.0)

Scout every leaf.Catch disease early.

Show it one leaf

The result will appear here

Scout every leaf.
Catch disease early.