How AI judges video games by their covers

Researchers used a multi-modal deep learning framework to train an artificial intelligence (AI) system to judge a video game by analysing its cover.

December 01, 2020 03:38 pm | Updated 03:50 pm IST

Researchers used both image-based and text-based models for generating genres of the video games.

Researchers used both image-based and text-based models for generating genres of the video games.

(Subscribe to our Today's Cache newsletter for a quick snapshot of top 5 tech stories. Click here to subscribe for free.)

The cover of a video game might not always be enough for gamers to know what the game is about. A pair of researchers seem to have developed a solution.

The duo used a multi-modal deep learning framework to train an artificial intelligence (AI) system to judge a video game by analysing its cover.

Yuhang Jiang and Lukun Zheng, in a research paper titled, ‘Deep learning for video game genre classification’, explained how they created a large training database and used it to train an AI system that classifies games from its cover.

“Video game genre classification based on its cover and textual description would be utterly beneficial to many modern identification, collocation, and retrieval systems,” researchers said.

They compiled a large dataset of 50,000 video games with their cover images, description text, title text, and genre information from a video game database. The database covered 21 genres, and they were labelled without considering internal relationships among some of the genres.

 

For instance, Real Time Strategy is a sub-genre of the genre Strategy. Similarly, Hack and slash/Beat’em up is a sub-genre of the genre Fighting. In such cases, researchers grouped genres like these as one genre and after alterations, they concluded with 15 different genres in total.

For simplification purposes, games with more than one genre were labelled under one genre. They used both image-based and text-based models for generating genres of the video games.

Additionally, a hybrid model was developed based on both images and texts.

They then trained the text and image recognition models with the compiled database. Researchers found that text-based models performed better than image-based ones, and the multi-modal outperformed the both models.

However, they also noted key problems in classifying the genres of the video game.

The researchers’ effort is not complete yet as there are a wide variety of video game genres that are not clearly defined. Covers vary in terms of colour, style, and textual information.

“In order to solve this task to a better level of performance, more efforts are needed on the creation of better data sets and the development of more sophisticated models,” researchers concluded.

Top News Today

Comments

Comments have to be in English, and in full sentences. They cannot be abusive or personal. Please abide by our community guidelines for posting your comments.

We have migrated to a new commenting platform. If you are already a registered user of The Hindu and logged in, you may continue to engage with our articles. If you do not have an account please register and login to post comments. Users can access their older comments by logging into their accounts on Vuukle.