app.py

import gradio as gr
import tensorflow as tf
from PIL import Image
import numpy as np
import json

# Load class labels
with open("class_indices.json", "r") as f:
    class_labels = json.load(f)

# Load model
model = tf.keras.models.load_model("best_model.h5")

def preprocess(img: Image.Image) -> np.ndarray:
    img = img.convert("RGB").resize((256, 256))
    img = np.array(img) / 255.0
    return np.expand_dims(img, axis=0)

def predict_disease(image: Image.Image) -> str:
    try:
        img = preprocess(image)
        preds = model.predict(img)
        idx = np.argmax(preds)
        label = class_labels[str(idx)]
        confidence = preds[0][idx]
        return f"### Prediction Result\n**Disease:** {label}\n**Confidence:** {confidence:.2%}"
    except Exception as e:
        return f"Prediction error: {e}"

css = """
.gradio-container {
    max-width: 600px;
    margin: auto;
    font-family: 'Segoe UI', Tahoma, Geneva, Verdana, sans-serif;
}
h1, h3 {
    color: #2c3e50;
}
.gr-button {
    background-color: #27ae60;
    border-radius: 5px;
    border: none;
    color: white;
    font-weight: bold;
}
.gr-button:hover {
    background-color: #219150;
}
"""

interface = gr.Interface(
    fn=predict_disease,
    inputs=gr.Image(type="pil", label="Upload a clear image of a tomato leaf"),
    outputs=gr.Markdown(label="Detection Result"),
    title="🍅 Tomato Leaf Disease Detector",
    description="Upload a tomato leaf image to detect diseases with confidence using a trained TensorFlow model.",
    allow_flagging="never",
    theme="default",
    css=css,
)

if __name__ == "__main__":
    interface.launch()



# ----------------------------------------------------------------------------------------------------------------------

# import gradio as gr
# import tensorflow as tf
# from PIL import Image
# import numpy as np
# import json

# # Load class indices from the uploaded file
# with open("class_indices.json", "r") as f:
#     class_indices = json.load(f)

# # Load model from local file
# model = tf.keras.models.load_model("best_model.h5")

# def preprocess_image(image):
#     # Resize to 256x256 as your model expects
#     image = image.convert("RGB").resize((256, 256))
#     img_array = np.array(image)
#     # Normalize pixel values between 0 and 1
#     img_array = img_array / 255.0
#     # Add batch dimension
#     img_array = np.expand_dims(img_array, axis=0)
#     return img_array

# def predict(image):
#     try:
#         img = preprocess_image(image)
#         preds = model.predict(img)
#         predicted_index = np.argmax(preds)
#         predicted_class = class_indices[str(predicted_index)]
#         confidence = preds[0][predicted_index]
#         return f"Predicted Disease: **{predicted_class}** with confidence {confidence:.2f}"
#     except Exception as e:
#         return f"Error: {str(e)}"

# iface = gr.Interface(
#     fn=predict,
#     inputs=gr.Image(type="pil", label="Upload Tomato Leaf Image"),
#     outputs=gr.Markdown(label="Prediction"),
#     title="Tomato Leaf Disease Detection",
#     description="Upload a tomato leaf image to detect the disease using a TensorFlow Keras model.",
# )

# if __name__ == "__main__":
#     iface.launch()