Landmark recognition using deep learning

Abstract

Large-scale picture retrieval is an important task in computer vision since it is linked to a variety of practical applications, such as object detection, visual place recognition, and product recognition. With applications in search, image understanding, apps, maps, medical, drones, and self-driving automobiles, computer vision has become omnipresent in our communities. Visual recognition tasks including image classification, localization, and detection are at the heart of many of these applications. Among all these practical applications, image classification for landmarks will be focus on this project. The solution for this project is to train the identical dataset with different classes to study the accuracy of the various model. In order to have better accuracy of the landmarks’ recognition, advanced algorithms are required to develop in order to train the model with big datasets. In this project, it aims to create a python-based model to classify landmark images with an appropriate label for model development and study how the different classes affect the result of the model. This can help a lot of people to recognize the landmark photos which has been taken and organize their photo collection with correct label. The report concludes with some discussion of the project's outcomes as well as suggestions for improvements.