LeafNet LeafNet

Tap to upload a leaf image

JPEG or PNG

Detection Settings

Confidence Threshold 0.33

Higher values show only strong detections.

Overlay Color

Light tones keep leaf details visible under overlays.

Overlays

Segmentation Masks
Bounding Boxes
Confidence Score

Explainability

Attention Heatmap

LeafNet

Diseased ROI Extraction

What it does

LeafNet detects and segments diseased regions on plant leaves. Capture or upload a leaf photo, and the model highlights affected areas with precision overlays — boxes, masks, and confidence scores.

Who it's for

Farmers monitoring crop health, plant pathologists conducting field surveys, and agricultural researchers analyzing disease spread. Works on any device, no specialised hardware required.

How to use

  • Tap the leaf area to upload an image
  • Press Run Diagnosis
  • Tap the eye icon to compare original vs result
  • Adjust threshold, color, and overlays in Settings
  • Enable Attention Heatmap to see where the model focused
  • Download the annotated result

Model

RF-DETR-Seg-Nano fine-tuned on a leaf disease segmentation dataset. Runs entirely on CPU via ONNX Runtime.