Computer Vision for Super Smash Bros Melee

This project runs a series of computer vision algorithms to analyze the following famous clip in which Moar Rzr Axe four stocks SilentWolf in less than at minute Evo 2014.

The analysis is fairly brittle and is not generalizable to other Super Smash Bros clips The dmg percentage detection and in game timer detection still need some work and do not work 100% of the time

Dependencies

• Numpy
• opencv (using opencv-contrib)
• pytesseract
• tesseract-OCR

Installation Instructions

Installing Dependencies

pip install numpy

pip install opencv-contrib-python

pip install pytesseract

Installing Tesseract-OCR

pytesseract is simply a wrapper for the CLI tool so you'll need to install Tesseract-OCR for your platform as well. After you are done make sure Tesseract-OCR is added to your PATH. You can confirm that everything works by running tesseract --version

For installation instructions see here

Running the program

After you have all the dependencies installed you should be able to run main.py. It will download the clip and start running the CV algorithms and print stats to the console.