Architecture.py

import tensorflow as tf
import cv2
import numpy as np
import os

# Function to load and preprocess an image
def preprocess_image(image_path):
    if os.path.exists(image_path):
        image = cv2.imread(image_path)
        image_rgb = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
        image_resized = cv2.resize(image_rgb, (224, 224))
        image_preprocessed = tf.keras.applications.mobilenet_v2.preprocess_input(image_resized)
        return image_preprocessed
    else:
        print(f"File {image_path} does not exist or path is incorrect.")
        return None

# Load the pre-trained MobileNetV2 model
model = tf.keras.applications.MobileNetV2(weights="imagenet")

# Function to predict landmarks and display results
def predict_landmarks(image_path):
    image = preprocess_image(image_path)
    if image is not None:
        # Make predictions using the model
        predictions = model.predict(np.expand_dims(image, axis=0))
        decoded_predictions = tf.keras.applications.mobilenet_v2.decode_predictions(predictions, top=5)[0]

        # Display the top 5 predicted landmarks
        for i, (imagenet_id, label, score) in enumerate(decoded_predictions):
            print(f"{i + 1}: {label} ({score:.2f})")

        # Display the image with annotations
        image_display = cv2.cvtColor(cv2.imread(image_path), cv2.COLOR_BGR2RGB)
        cv2.imshow("Landmark Recognition", image_display)
        cv2.waitKey(0)
        cv2.destroyAllWindows()

# Example usage
if __name__ == "__main__":
    image_path = "C:\\Users\\acer\\OneDrive\\Desktop\\Bnmit\\Projects & Assignments\\Landmark detection of a place\\t.jpg"
    predict_landmarks(image_path)