Skip to content
Switch branches/tags


Failed to load latest commit information.
Latest commit message
Commit time


This is a Python program that uses OpenCV to detect and identify playing cards from a PiCamera video feed on a Raspberry Pi. Check out the YouTube video that describes what it does and how it works:


Download this repository to a directory and run from that directory. Cards need to be placed on a dark background for the detector to work. Press 'q' to end the program.

The program was originally designed to run on a Raspberry Pi with a Linux OS, but it can also be run on Windows 7/8/10. To run on Windows, download and install Anaconda (, Python 3.6 version), launch Anaconda Prompt, and execute the program by launching IDLE (type "idle" and press ENTER in the prompt) and opening/running the file in IDLE. The Anaconda environment comes with the opencv and numpy packages installed, so you don't need to install those yourself. If you are running this on Windows, you will also need to change the program to use a USB camera, as described below.

The program allows you to use either a PiCamera or a USB camera. If using a USB camera, change line 38 in to:

videostream = VideoStream.VideoStream((IM_WIDTH,IM_HEIGHT),FRAME_RATE,2,0).start()

The card detector will work best if you use isolated rank and suit images generated from your own cards. To do this, run to take pictures of your cards. It will ask you to take a picture of an Ace, then a Two, and so on. Then, it will ask you to take a picture of one card from each of the suits (Spades, Diamonds, Clubs, Hearts). As you take pictures of the cards, the script will automatically isolate the rank or suit and save them in the Card_Imgs directory (overwriting the existing images).

Files contains the main script has classes and functions that are used by creates a video stream from the PiCamera, and is used by is a standalone script that can be used to isolate the rank and suit from a set of cards to create train images

Card_Imgs contains all the train images of the card ranks and suits


Python 3.6

OpenCV-Python 3.2.0 and numpy 1.8.2: See for how to build and install OpenCV-Python on the Raspberry Pi

picamera library:

sudo apt-get update
sudo apt-get install python-picamera python3-picamera


Python program that uses OpenCV to detect and identify playing cards from a PiCamera video feed on a Raspberry Pi



No releases published


No packages published