Session

Mechatronics, System Engineering and Robotics

Description

Image classification application has recently been covering a high number of research fields. In the other hand as the performance of the mobile devices is being updated day by day, the implementation of image recognition algorithms in them, is not only being trendy but very helpful in everyday tasks. With the automatic monument recognition, visiting a city is easy and fun. This application recognizes the captured monument, gives useful information and describes that particular landmark.

In this paper there are used four historical monuments of the city of Prizren, Kosovo and the aim is to classify these four famous monuments.

The images where taken specially for the research from the different angles of the city and the dataset for the classification process has been created.

One of the most convenient algorithms of Deep Learning is used for the classification of these four types of monuments. Using Matlab, a convolutional neural network architecture is adopted for the training of the created model.

After the model is fully trained, it is validated and tested to calculate the accuracy of the model.

Keywords:

Convolutional Neural Networks, monument recognition, deep learning, accuracy.

Session Chair

Peter Kopacek

Session Co-Chair

Thomas Duda

Proceedings Editor

Edmond Hajrizi

ISBN

978-9951-550-19-2

First Page

22

Last Page

29

Location

Pristina, Kosovo

Start Date

26-10-2019 11:00 AM

End Date

26-10-2019 12:30 PM

DOI

10.33107/ubt-ic.2019.149

Included in

Engineering Commons

Share

COinS
 
Oct 26th, 11:00 AM Oct 26th, 12:30 PM

IMAGE BASED RECOGNITION OF THE MONUMENTS IN PRIZREN

Pristina, Kosovo

Image classification application has recently been covering a high number of research fields. In the other hand as the performance of the mobile devices is being updated day by day, the implementation of image recognition algorithms in them, is not only being trendy but very helpful in everyday tasks. With the automatic monument recognition, visiting a city is easy and fun. This application recognizes the captured monument, gives useful information and describes that particular landmark.

In this paper there are used four historical monuments of the city of Prizren, Kosovo and the aim is to classify these four famous monuments.

The images where taken specially for the research from the different angles of the city and the dataset for the classification process has been created.

One of the most convenient algorithms of Deep Learning is used for the classification of these four types of monuments. Using Matlab, a convolutional neural network architecture is adopted for the training of the created model.

After the model is fully trained, it is validated and tested to calculate the accuracy of the model.