24f2004275/potato-disease-classifier

Potato Disease Classifier

CNN model for classifying potato leaf diseases: Early Blight, Late Blight, and Healthy.

Model Description

A Sequential CNN built with TensorFlow/Keras that classifies potato leaf images into three categories.

Architecture

• Input preprocessing: Resizing(256,256) + Rescaling(1/255)
• Data augmentation: RandomFlip + RandomRotation (training only)
• 6x Conv2D(32-64 filters, 3x3, ReLU) + MaxPooling2D(2x2) blocks
• Flatten + Dense(64, ReLU) + Dense(3, Softmax)
• Total params: 183,747

Classes

Class	Description
Potato___Early_blight	Early blight disease infection
Potato___Late_blight	Late blight disease infection
Potato___healthy	Healthy potato leaf

Performance

• Test Accuracy: ~99.6%
• Test Loss: ~0.009
• Validation Accuracy: ~99.5%
• Validation Loss: ~0.011

Usage

import keras
import numpy as np

model = keras.models.load_model("potato-disease-model.keras")

def predict(image_path):
    img = keras.utils.load_img(image_path, target_size=(256, 256))
    img_array = keras.utils.img_to_array(img)
    img_array = np.expand_dims(img_array, axis=0)
    predictions = model.predict(img_array, verbose=0)
    class_names = ["Potato___Early_blight", "Potato___Late_blight", "Potato___healthy"]
    predicted_class = class_names[np.argmax(predictions[0])]
    confidence = round(100 * float(np.max(predictions[0])), 2)
    return predicted_class, confidence