Karlo Dumbović

This project is a part of a wider project that focuses on interactive audience participation in live concerts, and in its original version, suggests the use of only one balloon as an interaction media. The goal of this project is to build a system that is able to estimate 3D world positions of several balloons during the concert, and to calculate them in real time. The system uses two static cameras placed in front of the audience. In every frame, it needs to detect each of the tracked balloons in both camera images. After detection, using their estimated 2D image position, it calculates their 3D world position. Once the system has calculated the positions, it sends them using the OSC protocol to a named IP address and port. The system is expected to be robust enough to work in different kinds of conditions that may occur during the live concert, such as fogginess and dynamic lighting. The system is made in C++ using the well-known computer vision library OpenCV. To see more, please visit https://goo.gl/eKXAJn.

The focus of this project is developing a system for detection and tracking of players in video recordings of football matches using computer vision techniques. The system is built to work on recordings that were filmed with a stationary camera, but there are also some useful approaches that would work well with a moving camera described. The three main parts of the project are field detection, player detection, and player tracking, among which the success of each one of them depends on the gain achieved by the ones previously listed. While building the system, there were many math and computer science algorithms and procedures used, such as particle and Kalman filters, k-d trees, an algorithm to find minimal edge cover in weighted graph, Hough transform, and k-means clustering. The system is built in C++ using OpenCV.

This project focuses on the development of a system for the pool game analysis. The analysis is carried out in two stages. At the beginning, a picture of a certain billiards situation is given to the system. The first stage is an image processing stage, in which a detection of pool game elements is carried out using OpenCV library. Pool game elements are playing surface and balls. The playing surface is detected as a rectangle defined by four lines. Balls are detected in two ways: as circles and as contours inside the playing surface. After they are detected, each ball goes through a recognizing stage. The recognizing stage uses certain parameters calculated from the ball picture and matches them to a base of previously recognized balls. After the image processing stage, the detected situation is ready to be analyzed. The situation is being displayed to the user as a 3D model of a table and balls. The user firstly chooses one ball and then one pocket in which he wants to pot the chosen ball. The system then using mathematics and relying on the pool game physics picks a set of shots that lead to the chosen pot. The user then selects one of the offered shots and then the system conducts a simulation of that shot in real time.

Mozgalo is the first and the only Croatian competition for students in data mining and big data.

Through their solutions, students gather useful knowledge and information among big amounts of structured and unstructured data. They use different statistical techniques, machine learning, and computer vision algorithms. One of Mozgalo’s primary goals is to encourage students to do analytical thinking and prediction modeling to independently develop their own creative solutions to real problems using their knowledge of computer science, statistics, and mathematics. In addition to that, students develop their communication and business skills, teamwork, and connections with companies from the IT and bank sectors. During the competition, educational workshops are being held and competitors are offered mentorship and online educational content. The competition is open to all students, in teams of 2 to 4 people.

Chatbot web application used as a helpful assistant for a charity organization. The app was previously developed in Python, suited for local use only. My task was to adapt it for a web application and deploy it to make it available for clients. I used Python and HTML/JavaScript to prepare the app for web deployment. I used Amazon Elastic Container Service (Amazon ECS), Amazon S3 (AWS S3), and Amazon CloudFront CDN for the deployment.