import gradio as gr
from transformers import pipeline
from huggingface_hub import hf_hub_download
import json

# Define model repository details
model_repo = "KhadijaAsehnoune12/LeafDiseaseDetector"
config_filename = "config.json"

# Download the config.json file from the model repository
config_path = hf_hub_download(repo_id=model_repo, filename=config_filename)

# Load the config.json file
with open(config_path, "r", encoding="utf-8") as f:
    config = json.load(f)

# Initialize the pipeline
pipe = pipeline(task="image-classification", model=model_repo)

# Define a custom prediction function
import numpy as np

# Define a custom prediction function
def predict(image):
    # Convert the image to numpy array if it's not already in that format
    if not isinstance(image, np.ndarray):
        raise ValueError("Input image must be a numpy array.")
    # Get the predictions from the pipeline
    predictions = pipe(image)
    # Get the predicted label index
    predicted_index = predictions[0]['label']
    # Map the index to the corresponding disease name using id2label
    label_name = config["id2label"][str(predicted_index)]
    # Optionally, you can add the confidence score as well
    confidence_score = predictions[0]['score']
    return f"{label_name} ({confidence_score:.2f})"


# Create Gradio interface
iface = gr.Interface(fn=predict,
                     inputs=gr.Image(type="numpy"),
                     outputs=gr.Textbox(),
                     title="Orange Disease Image Classification",
                     description="Detect diseases in orange leaves and fruits.",
                     examples=['MoucheB.jpg', 'verdissement.jpg'])

# Launch the app
iface.launch()