app.py

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

# Load the TFLite model
def load_tflite_model(model_path):
    interpreter = tf.lite.Interpreter(model_path=model_path)
    interpreter.allocate_tensors()
    return interpreter

# Preprocess image for TFLite model (assuming the model expects 224x224 RGB images)
def preprocess_image(image):
    image = image.resize((224, 224))  # Resize to model input size
    image = np.array(image).astype(np.int8)  # Convert to INT8
    image = np.expand_dims(image, axis=0)  # Add batch dimension
    return image

# Run inference with the TFLite model
def run_inference(interpreter, image):
    input_details = interpreter.get_input_details()
    output_details = interpreter.get_output_details()
    
    # Set input tensor
    interpreter.set_tensor(input_details[0]['index'], image)
    
    # Run inference
    interpreter.invoke()
    
    # Get the output tensor
    output_data = interpreter.get_tensor(output_details[0]['index'])
    
    return output_data

# Gradio interface function
def predict(image):
    interpreter = load_tflite_model("MNv2Flood_cat_Mar2025.tflite")  # Replace with your TFLite model path
    preprocessed_image = preprocess_image(image)
    prediction = run_inference(interpreter, preprocessed_image)
    
    # Get the predicted class by comparing the output values
    if prediction[0][0] > prediction[0][1]:
        return "Flood"
    else:
        return "No Flood"

# Create Gradio interface
interface = gr.Interface(
    fn=predict,  # Function that runs the model
    inputs=gr.Image(type="pil"),  # Input is an image, processed as PIL image
    outputs="text",  # Output is text indicating Flood or NoFlood
)

# Launch the app
interface.launch()