STEAMCODE - Community Detection on Steam Games

Team 7

Tian Runxin, 2016, Southern University of Science and Technology

Zheng Cuiyu, 2016, Zhejiang University

Jia Lianchen, 2017, Harbin Engineering University

Liu Yifan, 2017, Xi'an Jiaotong University

Team Leader: Tian Runxin

---

1. Fast go trough the project

2. How to Run

3. Result Visualization (using d3-3d framework[3])

Decription

Our project aims to investigate the similarity between header pictures (tiny posters) of games on the Steam[1] game platform, and find the relationship between header pictures, game categories and popularity, etc. For example, games belong to the same category may have similar header pictures.

Dataset & Subclasses

• Kind of data: Header (tiny poster) image of Steam games (460x215) in .jpg format.
• Source: The images are from Steam[1] and the data for further analysis is from Steamdb[2].
• Preprocessing:
  • It is needed to confirm that all the games have such a header image.
  • Normal image preprocessing for CNN.
  • Wash the data for further analysis.
• Additional data: It shown below in APIs part.

APIs

1. Game Header Poster

   

   Source: https://steamcdn-a.akamaihd.net/steam/apps/[APPID]/header.jpg

2. Game Detail (Additional data for analysis)

   https://steamdb.info/app/[APPID]/info/

   

3. Game tags (Data to confirm our expected results)

   https://steamdb.info/tags/

   

Reference

[1] Steam is a video game digital distribution platform developed by Valve Corporation.

[2] Steamdb is a third-party tool, which was made to give better insight into the applications and packages that Steam has in its database.

[3] https://github.com/Niekes/d3-3d

Poster

Deploy

To be continue...

If you really want to know the deployment method, please contact me.