Leaf image
Drop a leaf photo here
One clear photo of a single leaf, taken in even light, gives the model the most to work with.
JPG · JPEG · PNG
Select a leaf image first to enable analysis.

Disease Detection
Upload a leaf photograph and CropCare AI classifies it through the project's TensorFlow/Keras model. The prediction, its confidence and the class-wise probabilities you see are the model's own output.
Not evaluated yet— when no model is configured
No prediction is invented or simulated. If the trained weights are missing, this page says so and explains how to configure the model instead of guessing.
Drop a leaf photo here
One clear photo of a single leaf, taken in even light, gives the model the most to work with.
JPG · JPEG · PNG
Select a leaf image first to enable analysis.
Not evaluated yet
Select a leaf image on the left to see the model result here.
Not evaluated yet
03Reading the Result
The result panel reports what the model inferred from your leaf image. Interpret it as one piece of evidence, alongside what you can see on the plant.
General background on how the predicted class was learned and how far the model can be trusted.
Not evaluated yet.Disease-specific notes appear with a real prediction from the trained model; no notes are shown before then.
Disease patterns
Visual symptoms such as spots, lesions, yellowing and wilting are what the CNN's learned filters respond to.
Model limits
The prediction reflects the classes in the training dataset. Use the class list to check which diseases the model can actually name.
Practical follow-ups that apply after any result, useful for the demonstration and viva walkthrough.
Inspect the leaf under natural light and compare its symptoms with the predicted class.
Keeping suspect plants apart from healthy ones limits the chance of spread while you confirm the diagnosis.
Shoot one clear leaf against a plain background. Blurry or cluttered photos lower the model result quality.
For treatment decisions, confirm with an agricultural extension officer or a plant pathologist.

01Deep learning project · leaf diagnosis
Deep learning reads a leaf image and classifies its disease. You see the model result and the next steps.
Not evaluated yet
Not evaluated yet
Not evaluated yet
Not evaluated yet
02How it works
No. 03 — Objectives & Benefits
Crop diseases take hold quietly. A lesion that is obvious on a photograph can be easy to miss in a field, and late detection usually means lost yield that cannot be recovered. CropCare AI is a Deep Learning study of how far an image classifier can be pushed in reading a leaf and naming what it shows.
The first objective is technical: build and train convolutional models that take a single leaf image and classify its disease, using a consistent preprocessing pipeline between training and inference. The second is developmental: establish a clean workflow across dataset preparation, training experiments, evaluation and error review, so that the resulting model is measured honestly rather than assumed to work.
Alongside the model sits an examinable record — training curves, per-class precision, recall and F1-scores, and a confusion matrix that shows exactly where the classifier is confident and where it is not. Every value on this site is read from stored experiment output; anything not yet measured is shown as Not evaluated yet.
For a student and their evaluator, the benefit is a working, traceable pipeline: from image upload through analysis to a predicted class and recommended next steps, with the model's reasoning exposed as class-wise probabilities. For the wider goal of early disease identification, a model that flags a likely problem early can support a more timely response — providing its limits are understood as clearly as its strengths.
No comments yet. Be the first!