HerbaScanPlant ID & Herbal Guidance

Module 1

Plant identification

Upload a clear photo of a leaf or plant. The image is validated here, then classified by the trained MobileNetV2 model running in the project's analysis service, which also returns a Grad-CAM explanation.

Live model inference

1. Upload image

JPG, JPEG or PNG up to 8 MB. The photo is sent to the analysis service for classification.

Drag and drop a leaf or plant photo

JPG, JPEG or PNG · up to 8 MB · best results with a single clear leaf

Preprocessing applied by the service

Target size: 224 × 224 px (MobileNetV2 input)

  1. Decode JPG/PNG and convert to RGB
  2. Resize to 224 × 224
  3. Apply MobileNetV2 preprocess_input (scales to [-1, 1])
  4. Batch as a single tensor of shape (1, 224, 224, 3)

Inference call: POST /api/predict-image (multipart image → class label, confidence and Grad-CAM).

Identification result

No result yet

Choose an image and select “Identify plant”. Your result card will show the predicted plant, the model's confidence, the Grad-CAM heatmap, botany, medicinal properties, traditional uses and precautions.