International Research Journal of Engineering and Technology (IRJET) Volume: 12 Issue: 11 | Nov 2025
www.irjet.net
e-ISSN: 2395-0056 p-ISSN: 2395-0072
LANDMARK RECOGNITION USING CNN MODEL Anusha P Neeralakeri¹, Chandrika K Math², Kangoni Mayuri Sunil³, Prof.Parimala. R. Muttannavar4 1,2,3, Student , Department of Information Science and Engineering , Basaveshwar Engineering College, Bagalkote,
India 4Assistant Professor,Department of Information Science and Engineering, Basaveshwar EngineeringCollege,
Bagalkote, India. -----------------------------------------------------------------------***-------------------------------------------------------------------------Abstract: The increasing availability of digital images from tourists, travelers, and mapping platforms has created a demand
for intelligent systems capable of automatically identifying famous landmarks. This paper presents a Landmark Recognition System that uses a Convolutional Neural Network (CNN) model to classify landmarks from user-uploaded images. The system leverages a pre-trained TensorFlow Hub model optimized for Asian landmarks and integrates a Streamlit web interface to provide real-time predictions and user interaction. Once an image is processed, the system identifies the landmark and retrieves its geographical details—including address, latitude, and longitude—through the Geopy library, displaying them interactively on a map. The proposed system achieves high prediction accuracy and offers an intuitive interface for end users. This work demonstrates how the integration of deep learning and geolocation technologies can enhance applications in tourism, navigation, and education.
Keywords: Convolutional Neural Network (CNN), Landmark Recognition, Deep Learning, TensorFlow Hub, Streamlit, Image Classification, Geolocation, Artificial Intelligence, Tourism Applications
1. INTRODUCTION Landmarks serve as important cultural, historical, and geographical markers that define the identity of regions and attract global interest. With the exponential growth of digital imagery—particularly through social media, travel platforms, and mapping services—there is a rising demand for automated landmark recognition systems. Manual identification of landmarks from such massive image datasets is time-consuming and prone to errors. Therefore, developing intelligent systems that can recognize landmarks automatically has become crucial for applications in tourism, navigation, cultural preservation, and geographic information systems. In recent years, several approaches have been proposed to address this problem. Early methods relied on traditional computer vision techniques such as Scale Invariant Feature Transform (SIFT) and Speeded-Up Robust Features (SURF) to extract and match features between images. However, these methods required extensive manual feature engineering and performed poorly under varying lighting conditions, viewing angles, and partial occlusions. Subsequent studies began exploring machine learning techniques and geotagged datasets for image-based localization. For instance, Heys and Efros (2008) attempted geographic information estimation using nearest neighbour searches, while Weyand et al. (2016) introduced PlaNet, a CNN-based model trained on millions of geotagged images to predict photo locations. Although these models marked major progress, they often required massive computational resources, large datasets, and lacked real-time performance.
Figure 1.Landmark Recognition System
© 2025, IRJET
|
Impact Factor value: 8.315
|
ISO 9001:2008 Certified Journal
| Page 618