BilliardEye

A sport video analysis software for billiard matches.

To develop the system, the following tasks were solved:

• Table segmentation
• Table sides recognition
• Balls localization and segmentatation
• Balls identification
• Balls tracking
• Map rendering
• System performance evaluation

Full details on the project specifics and development, system outputs and program execution can be found in the report.

Example of system output:

game4_clip2.mp4

---

Execution instructions

Project building:

git clone https://github.com/mattia-toffolon/BilliardEye.git
cd BilliardEye
mkdir build
cd build
cmake ..
make

System setup:

cd ..
unzip samples.zip

Program execution:

cd build
./outMain "path-to-video" "path-to-first-frame" "path-to-output-dir"
./perfMain "path-to-samples-folder" "number-of-samples"

NOTE: The system is based on OpenCV. To succesfully execute the program, such library shall be correctly installed.