app.py

import gradio as gr 
from tensorflow.keras.models import load_model 
from tensorflow.keras.preprocessing.image import img_to_array 
from PIL import Image 
import numpy as np 
import requests 
 
# Load your trained .h5 model 
model = load_model("model.h5") 
 
# Hugging Face LLM (e.g., Mistral) 
HF_TOKEN = "hf_your_token_here" 
LLM_API = 
"https://api-inference.huggingface.co/models/mistralai/Mistral-7B-In
 struct-v0.1" 
headers = {"Authorization": f"Bearer {HF_TOKEN}"} 
 
# Prediction and explanation function 
def classify_and_explain(image): 
    image = image.resize((225, 225)) 
    img_array = img_to_array(image) / 255.0 
    img_array = np.expand_dims(img_array, axis=0) 
 
    prediction = model.predict(img_array) 
    predicted_class = int(np.argmax(prediction, axis=1)[0]) 
    class_label = f"Class_{predicted_class}"  # Or use actual class 
names if available 
 
    # Get explanation from LLM 
    prompt = f"Explain the plant disease {class_label} and how to 
treat it." 
    response = requests.post(LLM_API, headers=headers, 
json={"inputs": prompt}) 
    llm_text = response.json()[0]['generated_text'] if 
isinstance(response.json(), list) else response.json() 
 
    return class_label, llm_text 
 
# Gradio UI 
interface = gr.Interface( 
    fn=classify_and_explain, 
    inputs=gr.Image(type="pil"), 
    outputs=["text", "text"], 
    title="Plant Disease Predictor & Explainer", 
description="Upload a leaf image to detect plant disease and get 
treatment advice using LLM" 
) 
interface.launch()