Rubiks-Scanner

A prototype computer vision application to replace human data entry at Rubik's Cube competitions.

Created by Bridget Andersen, Samantha Reid, and Steven Stetzler.

For an in-depth description of our project, read our write-up contained in rubiks_scanner_info.pdf

Software Requirements:

• Python 3.5.3
• OpenCV 3.2.0
• Tensorflow 1.0.0
• Keras 2.0.2

Run one of either of the following in order to test our project.

• rubiks_scanner_phone_video.py
• rubiks_scanner_static_image.py
• rubiks_scanner_webcam_video.py

The static_image version uses a saved image of a scorecard and matches it to a template. The webcam_video version uses the output of the current computer's default webcam as a video capture device. The phone_video version connects to an IP camera at a specified IP address, hardcoded into that file.

Note on Windows/OS X/Linux compatibility: Some file loctions are hardcoded using Windows path syntax in our files. Make changes at the following locations in order to allow the code to run on OS X/Linux

• rubiks_scanner_core.py: line 72
• rubiks_scanner_phone_video.py: line 15
• rubiks_scanner_static_image.py: line 6, 7
• rubiks_scanner_webcam_video.py: line 6

You can find a database storing information from all processed scorecards at https://rubiksscanner.firebaseio.com/

If you need access to view the database, send a request to [email protected]

References:

Many sources and tutorials were used in the creation of this project. You can find them at the following locations

1. SIFT Feature Matching
2. Rectangle Contour Extraction
3. OpenCV Video Capture