This repository contains Python code to retrieve Steam games with similar store banners, using OpenAI's CLIP.

Image similarity is assessed by the cosine similarity between image features encoded by CLIP.

• Install the latest version of Python 3.X.
• Install the required packages:

python -m pip install --upgrade pip
pip install -r requirements.txt

CLIP is a neural network:

• combining an image encoder (either ResNet or Vision Transformer ViT) and a text encoder (Transformer),
• pre-trained on the WebImageText (WIT) dataset, consisting of 400 million (image, text) pairs.

In this repository, the image encoder is ViT-B/32.

Data is available in download-steam-banners-data/.

The most recent data snapshot was downloaded with this Colab notebook on January 9, 2021.

This snapshot is shared as an archive (original_vertical_steam_banners.tar, 1.5 GB) on Google Drive.

It consists of vertical Steam banners (300x450 resolution), available for 29982 out of 48792 games, i.e. 61.4% of games.

Resized images are provided in the same repository for resolutions 256, 224, 128, 64, etc.

The list of appIDs (before any potential filtering) is from steam-store-snapshots.

Information is also provided in .txt logs about a possible filtering out of images based on:

• image size (before resizing images):
  • there is 1 image with resolution 600x900,
  • this is not a big issue as the image ratio is equal to the expected ratio for 300x450 images,
• image channels (before and after resizing images):
  • most images are 'RGB' (for true color images) ; total: 29642 images,
  • a few images are 'L' ('luminance' for greyscale images) ; total: 306 images,
  • very few images are 'CMYK' (for pre-press images) ; total: 34 images,
• blank images:
  • there are 2 images either totally black (appID: 603280) or totally white (appID: 1076060),
  • these specific images are not reported about since they already appear in the log w.r.t. image channels.

It is up to the reader to filter out the dataset based on these logs. Logs can be reproduced with this Colab notebook.

Run match_steam_banners_with_CLIP.ipynb.

This will:

• compute and store the 512 features corresponding to each banner,
• find the 10 most similar store banners to curated query appIDs,
• find the one most similar store banner to all appIDs available on the store, then display the most unique games.

NB: by default, query appIDs consist of:

• the top 100 most played games during the past two weeks, according to SteamSpy,
• a few manually curated games.

NB: unique games are ones which are the most dissimilar (low similarity score) to others to their first neighbor.

Results can be interactively explored with web apps:

• using an appID,
• using a dropdown menu.

• Caveat
• Similar games
  • the same franchise
  • A common word
• Unique games

The CLIP embedding for the ~30k banners is shared on Google Drive.

Results obtained with OpenAI's CLIP are shown on the Wiki.

The linked pages contain a lot of images and might be slow to load depending on your Internet bandwidth.

Direct links to similarity results are available below:

• for each game, find the 10 most similar games.

For instance:

Cities

Dinosaurs

WWII / Heroes / War / W

The model is able to retrieve games from the same franchise.

Arma / Army / Arms

Borderlands / Land

Braveland / Capital "B" letter

Call of Duty

Age of Empires / Empire(s) / Emperor

Fallout

Grand Theft Auto

Guacamelee

Hitman

λ / Half-Life / Hal

NBA / 2K

The model relies on words present in the images, as shown in the following examples.

Forest

Legend

Civilization / Civ / tion

Dawn

Dead / Light

Don't / Together

Heroes

Hunt

Life / Strange

Monster / Hunt

Planet

Portal

Rain

Secret / Lab

Story

Story (again)

Truck

Black

Ust

Dead

Valley

The absence of words allows to retrieve similar image content.

Direct links to similarity results are available below:

• for each unique game, display the 1 most similar game,
• a grid of unique games.

For instance:

• Google's ViT (Vision Transformer):
  • Official Github repository
  • Official Colab notebook
  • Dosovitskiy, Alexey, et al. An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale. 2020.
• Open AI's CLIP (Contrastive Language-Image Pre-Training):
  • Official blog post
  • Official Github repository
  • Official Colab notebook
  • Radford, Alec, et al. Learning Transferable Visual Models From Natural Language Supervision. arXiv 2021.
• My usage of CLIP:
  • steam-CLIP: retrieve games with similar banners, using OpenAI's CLIP (resolution 224),
  • steam-image-search: retrieve games using natural language queries,
  • heroku-flask-api: serve the matching results through an API built with Flask on Heroku,
  • heroku-clip: deploy CLIP on Heroku,
• MobileNet v3:
  • match-steam-banners: retrieve games with similar banners, using MobileNet v3 (resolution 256),
• MobileNet v1:
  • download-steam-banners: retrieve games with similar banners, using MobileNet v1 (resolution 128),
  • download-steam-screenshots: retrieve games with similar screenshots, using MobileNet v1 (resolution 128).