app.py

import gradio as gr
import tensorflow as tf
from huggingface_hub import hf_hub_download
from PIL import Image
import numpy as np

# Load model from Hugging Face Hub
model_path = hf_hub_download(repo_id="arpitsharrrma/soilnet-model", filename="SoilNet.keras")
model = tf.keras.models.load_model(model_path)

# Define class labels (adjust if needed)
class_names = ['Alluvial Soil', 'Black Soil', 'Clay Soil', 'Red Soil', 'Sandy Soil']

def predict_soil(image):
    image = image.resize((150, 150))  # Model input size
    img_array = np.array(image) / 255.0
    img_array = img_array.reshape(1, 150, 150, 3)
    predictions = model.predict(img_array)
    predicted_class = class_names[np.argmax(predictions)]
    confidence = float(np.max(predictions)) * 100
    return f"{predicted_class} ({confidence:.2f}% confidence)"

# Gradio Interface
interface = gr.Interface(
    fn=predict_soil,
    inputs=gr.Image(type="pil", label="Upload Soil Image"),
    outputs=gr.Textbox(label="Predicted Soil Type"),
    title="SoilNet - Soil Type Classifier",
    description="Upload a soil image and the model will predict the soil type using deep learning."
)

interface.launch()