app.py

import os
import openai
import streamlit as st
from PIL import Image
from tensorflow.keras.models import load_model
from tensorflow.keras.preprocessing import image
import numpy as np

# Configuration
UPLOAD_FOLDER = 'static/uploads'
ALLOWED_EXTENSIONS = {'png', 'jpg', 'jpeg'}
TARGET_SIZE = (256, 256)

# OpenAI API Key (make sure to keep this key secure and not expose it in public repositories)
OPENAI_API_KEY = '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'
openai.api_key = OPENAI_API_KEY

# Ensure the upload folder exists
os.makedirs(UPLOAD_FOLDER, exist_ok=True)

# Load trained model
MODEL_PATH = 'desnet2.h5'
model = load_model(MODEL_PATH)

classes = [
    'American Bollworm on Cotton', 'Anthracnose on Cotton', 'Aphids', 'Army worm', 'Carpetweeds', 'Crabgrass', 'Eclipta',
    'Flag Smut', 'Goosegrass', 'Healthy', 'Leaf Curl', 'Leaf smut', 'Morningglory', 'Mosaic sugarcane', 'Nutsedge', 'PalmerAmaranth',
    'Powdery_Mildew', 'Prickly Sida', 'Purslane', 'Ragweed', 'RedRot sugarcane', 'RedRust sugarcane', 'Rice Blast', 'Sicklepod',
    'SpottedSpurge', 'SpurredAnoda', 'Sugarcane Healthy', 'Swinecress', 'Target_spot', 'Tungro', 'Waterhemp', 'Wheat Brown leaf Rust',
    'Wheat Stem fly', 'Wheat aphid', 'Wheat black rust', 'Wheat leaf blight', 'Wheat mite', 'Wheat powdery mildew', 'Wheat scab',
    'Wheat___Yellow_Rust', 'Wilt', 'Yellow Rust Sugarcane', 'bacterial_blight in Cotton', 'bollrot on Cotton', 'bollworm on Cotton',
    'cotton mealy bug', 'cotton whitefly', 'curl_virus', 'fussarium_wilt', 'maize ear rot', 'maize fall armyworm', 'maize stem borer',
    'pink bollworm in cotton', 'red cotton bug', 'thirps on cotton'
]

def allowed_file(filename):
    return '.' in filename and filename.rsplit('.', 1)[1].lower() in ALLOWED_EXTENSIONS

def predict_disease(image_path, model):
    test_image = image.load_img(image_path, target_size=TARGET_SIZE)
    test_image = image.img_to_array(test_image)
    test_image = test_image / 255
    test_image = np.expand_dims(test_image, axis=0)
    result = model.predict(test_image)
    result = result.ravel()
    index = np.argmax(result)
    pred = str(classes[index])
    return pred

def get_disease_info(disease_name):
    response = openai.Completion.create(
        engine="text-davinci-003",
        prompt=f"Explain the cause and solution for the following cotton plant disease: {disease_name}",
        max_tokens=150
    )
    return response.choices[0].text.strip()

# Streamlit app
st.title("Cotton Disease Detection")
st.write("Upload an image to detect the disease.")

uploaded_file = st.file_uploader("Choose an image...", type=["png", "jpg", "jpeg"])

if uploaded_file is not None:
    if allowed_file(uploaded_file.name):
        # Save the uploaded file to the upload folder
        image_path = os.path.join(UPLOAD_FOLDER, uploaded_file.name)
        with open(image_path, "wb") as f:
            f.write(uploaded_file.getbuffer())

        # Resize the image
        img = Image.open(image_path)
        img = img.resize(TARGET_SIZE)
        img.save(image_path)

        # Predict the disease
        prediction = predict_disease(image_path, model)
        
        # Display the uploaded image
        st.image(image_path, caption='Uploaded Image.', use_column_width=True)
        st.write(f"Prediction: {prediction}")

        # Get disease information from OpenAI GPT
        disease_info = get_disease_info(prediction)
        st.write("Disease Information:")
        st.write(disease_info)
    else:
        st.write("Please upload an image file (png, jpg, jpeg).")
else:
    st.write("No file uploaded.")