angela-sha / st-2019

Defect Prediction Code and Software Testing Work

ST-2019

Automated Glitch Detection in Video Games

Description:

This work includes generating corrupted images by inserting several type of artifacts to images/videos and using the "ensemble" model to classify normal and corrupted images.

The program to generate corrupted images is under the "Glitchify" folder.

The ensemble model to classify images as glitched or not is under the "Ensemble" folder.

Methods to extract features that we have explored but not used in the ensemble model are included in the "Experiment" folder.

Some old tools and outdated models are in the "Archive" folder.

Setup:

Run the following two commands to install the required packages, or to install all the packages listed in "setup.txt".

chmod +x install_packages.sh

bash install_packages.sh

Contributers:

Parmida Davarmanesh, University of Michigan

Kuanhao Jiang, University of Pennsylvania

Tingting Ou, Johns Hopkins University

Artem Vysogorets, University of Massachusetts Amherst

Shantanu Joshi, University of California, Los Angeles

Nicholas Malaya, AMD Research