Dice Detection

A Computer Vision project. Documentation Link

Developed by:

• Arthur Chieppe
• Luiza Valezim
• Vinícius Eller

An API developed with the goal to use Computer Vision and Machine Learning in order to identify and sum the values of the total dices.

Project demo: https://youtu.be/54zDDRuU2X8

There are two installation methods:

Using pip

 pip install cv-dice-detection

PyPi reference: https://pypi.org/project/cv-dice-detection/

Using the repository

1. Clone the repo

    git clone https://github.com/Vigrel/dice-detection.git

2. Create and activate virtualenv

    cd dice-detection/
    python3 -m virtualenv .venv
    source .venv/bin/activate

3. Run a demo

   python3 src/dice_detection_demo.py

Configuring the Environment

To ensure dice detection accuracy, set up the environment as follows:

1. Surface: Place a white piece of paper on the table for dice rolling. The white background enhances contrast for better detection.

2. Camera Position: Position the camera in a top-down view, parallel to the table surface. The camera height should be 40 centimeters above the table for a clear and consistent perspective.

3. Adjust CAMERA_DISTANCE: In the configuration module, adjust the CAMERA_DISTANCE parameter to reflect the actual distance in centimeters between the camera and the table surface. This parameter is crucial for accurate calculations and reliable dice number detection.

4. Camera Calibration (if necessary): Calibrate the camera if needed for accurate measurements. Refer to your camera documentation for calibration procedures.

Follow these guidelines to create an environment for effective dice detection, ensuring optimal system performance.

References:

• Golsteyn, Quentin - https://golsteyn.com/writing/dice