Multi-Camera Soccer Tracker

We developed a system for the tracking of players and ball in a soccer video. Work done at the Advanced Digital Sciences Center under Stefan Winkler. Details can be found in the attached report.

Contents

• Requirements
• Brief Project Structure
• Results
• Usage

Requirements

What we used to develop the system

• OpenCV 3.3
• Microsoft Visual Studio Community 2015
• Windows 10

Brief Project Structure

├── Soccer Tracker                 : Directory containing the source and header files of the project
├── dataset                        : Directory containing the dataset and the results folder
├── others                         : Directory containing images for instructions
├── Soccer Tracker.sln             : MSVC Project Solution
|── Report.pdf                     : Report

Results

Usage

• Install OpenCV

• Download the dataset at http://www.issia.cnr.it/wp/dataset-cnr-fig/ and place them in the dataset directory

  

• Launch Soccer Tracker.sln.

• Set solution configuration and platform to Release and x64.

• Set project properties (see images below) if using a different version of OpenCV.

• Enable OpenMP support

• Build the solution
• Ensure that the files opencv_ffmpeg330_64.dll and opencv_world330.dll are in the release directory (created only after building the solution). Both files can be found at <your_opencv_folder>\build\bin.

• Run the program via the Start Without Debugging option (Ctrl + F5)
• The video will be saved in dataset/results