Pokémon TCG Search Engine: Use AI to Catch Them All

Trading Card Games (TCGs) have been a source of joy and excitement for millions of people. For me, it all began in elementary school when I spent countless hours playing Pokémon TCG with my friends. Pokémon isn’t just a game; it is a cultural phenomenon, bringing friends together and sparking endless debates over strategies and collections.

In our office, everyone has their own story. Some grew up with Pokémon, while others were enchanted by Magic: The Gathering and its breathtaking fantasy artwork. Collecting sports cards was also a big part of childhood for many, especially in my country, where basketball and ice hockey cards were immensely popular. Cards are still loved, played, collected, invested and traded by geeks, collectors, and sports fans across the world! Their market is growing, and so is the need for automation of image processing and identification on websites and apps for collectors.

That’s why we created an online search system for collectible card items, whether they are Trading Card Games or Sports Cards. Our solution can be built and deployed for your cards in just a matter of days. At the core of this system is a powerful card identification engine powered by AI and accessible via a REST API. Simply send an image of a card from your app, and the API will return detailed information about it—quickly and accurately. This seamless integration is designed to make scanning, managing, documenting, and cataloging trading cards effortless.

Where can you use visual AI for cards?

Trading card games (TCGs, orトレーディングカード) can consist of tens of thousands of cards. In principle, building a basic image classifier (TCG categorization models) based solely on image recognition leads to low precision and is simply not enough for more complicated problems.

However, we are able to build a complex similarity system that can recognize, categorize, and find similar cards by a picture. Once trained properly, it can deal with enormous databases of images it never encountered before. With this system, you can find all the information, such as the card title, set, set code, year of release, series, or values of submissions on eBay.

Tip: Check out our Computer Vision Platform to learn about how basic image recognition systems work. If you are not sure how to develop your card search system, just contact us and we will help you.

Collectibles are a big business and some cards are really expensive nowadays. Who knows, maybe you have the card of Charizard or Kobe Bryant hidden in your old box in the attic. We can develop a system for you that can automatically analyze the bulk of trading cards sent from your customers or integrate it into your mobile/smartphone app.

ONE API TO RULE THEM ALL

AI That Understands Collectibles

Recognize, analyze & grade collectibles, find them in databases and marketplaces. Our API was build by collectors for collectors.

Yes! / はい！

What can visual search do for the trading cards market?

In the last year, we have been building a universal system able to train visual models with numerous applications in image search engines. We already offer visual search services for photo search. But, they are optimized mostly for general and fashion images. This system can be tuned to trading cards, coins, furniture & home decor, arts, and real estate, … there are infinite use cases.

In the last decades, we have all witnessed the growth of the TCG community. However, technologies based on artificial intelligence have not yet been used in this market. Plus, even though the first system for scanning trading cards was released by ebay.com, it was not made available for small shops as an API. And since trading card games and visual AI are a perfect match, we are going to change it – with a card image search.

Tip: Check out Visual Product Search to learn more about visual search applications.

Which TCG cards could visual AI help with?

An image search engine is a great approach when the number of classes for the image classification is high (above 1,000+). With TCGs, each card represents a unique class. A convolutional neural network (CNN) trained as a classifier can have poor results when working with a larger number of classes.

Pokémon TCG contains more than 20,000 cards (classes), Magic the Gathering (MTG) over 90.000, and the same goes for basketball or any other sports cards. So basically, we can build a visual search system for both:

• Trading card games (Magic the Gathering, Lord of the Rings, Pokémon, Yu-Gi-Oh!, One Piece, Warhammer, and so on)
• Collectible sports cards (like Ice Hockey, Football, Soccer, Baseball, Basketball, MMA/UFC, and more)

A visual search/recognition technology is starting to be used on E-bay when listing trading and sports cards for sale. However, this is only available in the e-bay app on smartphones. The app has a built-in scanning tool for cards and can find the average price with additional info.

Our service for card image search can be integrated into your website or application. And you can simply connect via API through a smartphone, computer, or sorting machine to find exact cards by photo, saving a lot of time and improving the user experience!

Our Pokémon Card Image Search

We’ve been recently training an AI (neural network) model for Pokémon trading cards, Yugioh!, Magic The Gathering, One Piece, Lorcana, Dragon Ball and other games. Why these? Pokémon is the most played TCG in the world, the game has simple rules and an enormous fan base. Very popular are also MTG and Yugioh! Some cards are really expensive, but more importantly, they are traded heavily!

With this model, we built a reverse search for finding the exact Pokémon card, MTG, Yugioh! and other games, which achieved 97%+ accuracy (i.e. exact image match). And we are still continuously working on new features and improvements. Our goal is to have the best and the most precise system based on visual AI. This search system can return you the edition of the card, language, name of the card, year of release and much more.

If you would like to try the system on these three trading card games, then the endpoint for card identification (/v2/tcg_id) from the Collectibles Recognition service is the right choice for you. If you need to tune it on your image collections or have any other games or cards (sports) then just contact us and we can build a similar service for you.

Automatic grading and inspection of cards with AI

A lot of companies are grading sports & trading cards manually. Our visual AI can be trained to detect corner types, scratches, surface wear, light staining, creases, focus, and borders. The Image recognition models are able to identify marks, centering errors, scratches, damaged corners & edges, print defects and other special attributes.

For example, PSA is a company that has developed its own grading standards for automatic card grading (MINT). With our platform and team, you can automatize the entire workflow of grading with just one photo. We provide several solutions for computing card grades and card condition.

Customized solution for visual search

With the new custom similarity service, we are able to create a custom solution for trading card image search in a matter of weeks. The process for developing it is quite simple:

1. We will schedule a call and talk about your goals. We will agree on how we will obtain the training data that are necessary to train your custom machine-learning model for the search engine.
2. Our machine-learning specialists will assemble a testable image search collection and train a custom machine-learning model for you in a matter of weeks.
3. After meeting all the requirements of PoC, we will deploy the system to production, and you can connect to it via Rest API.

Image Recognition of Collectibles

Machine learning models bring endless possibilities not only to pop culture geeks and collectors, but to all fields and industries. From personalized recommendations in custom fashion search engines to automatic detection of slight differences in surface materials, the visual AI gets better and smarter every day, making routine tasks a matter of milliseconds. That is one of the reasons why it is an unlimited resource of curiosity, challenges, and joy for us, including being geeks – professionally :).

Ximilar is currently releasing on a ready-to-use computer vision service able to recognize collectibles such as TCG cards, coins, banknotes or post stamps, detect their features and categorize them. Let us know if you’d like to implement it on your website!

If you are interested in a customized AI solution for collector items write us an email and we will get back to you as soon as possible. If you would like to identify cards with our Collectibles Recognition service just sign up via app.ximilar.com.