dice-recognition

2024, Till Bovermann

recognise and count dice eyes

Todo

• add ROI selection and cropping

Installation

best to install into a virtual environment

set up a virtual environment

python -m venv .venv
source .venv/bin/activate

install the required packages

python -m pip install -r requirements.txt

Usage

• Stable light and background conditions are important.
• A setup with the camera and a diffused LED light source below a glass plate worked well.
• A translucent white sheet of paper on top of the glass removes the background and isolates the dice.
• Adjusting the camera position to be in a slight angle helps to reduce reflections.

single image

Requires a single image of a dice with a white background. The script will detect the dice eyes and print the result.

python getDiceEyes.py --help

run an example

python getDiceEyes.py --min 0.3 --max 0.9 data/dice_5_eyes_4.jpg

video

Use a video source to detect dice eyes. The script will open a window and display the video stream. The dice eyes will be highlighted in the video stream.

For best results, adjust the parameters --min and --max and set the zoom level.

python videoGrabber.py --window --src 0 --fps 1 --min 0.025 --dilateIterations 2 --kernelSize 1 --max 0.08 2>/dev/null