Data used in the paper Automatically Detecting Visual Bugs in HTML5 Games

This repository contains the snapshots (i.e., screenshot, object representation pairs) and game assets collected from our test game for use in the paper Automatically Detecting Visual Bugs in HTML5 Games, accepted at ASE 2022.

The data in this repository can be used to benchmark new visual bug detection approaches for HTML5 games. The source code for our test game and our 24 synthetic visual bugs can be found at the following link.

Project page

Directory structure

data/
  | - assets/
  | - exp_0/
  | | - a/
  | | | - 0.png
  | | | - 0.json
  | | | ...
  | | | - 9.png
  | | | - 9.json
| | ...
  | | - j/
| - exp_appearance_1
  | ...
  | - exp_state_6

Data description

assets/
Source images from the test game

exp_0/
Snapshots with no visual bugs injected (okay/non-buggy experiment)

exp_*_{1,2,3,4,5,6}/
Snapshots for each of the injected visual bugs (buggy experiments)

*/{a,b,c,d,e,f,g,h,i,j}/
10 runs of data for each experiment

*/{0,1,2,3,4,5,6,7,8,9}.png
10 snapshots per experiment, each snapshot has a screenshot (png)

*/{0,1,2,3,4,5,6,7,8,9}.json
10 snapshots per experiment, each snapshot has a COR1 (json)

1COR = objects representation