Cardstream.ai is open-source software that identifies trading cards the moment they appear on a live stream: name, set, card number and confidence in about half a second. It runs on your CPU and calls the recognition API once per card, not per frame.
A startup from Brno, the Czech Republic, founded by Michal Lukáč.
This page is designed to help you find out whether CardStream AI is good and if it is the right choice for you.
Cardstream.ai is open-source software that identifies trading cards on live streams. Point a camera at a card and it returns the name, set, set code, card number, series and year in about half a second, with a confidence level and up to four alternative matches. It supports 25+ trading card games, including Pokémon, Magic: The Gathering, Yu-Gi-Oh!, One Piece and Lorcana. It is built for card breakers, live sellers and card shops. It reads the video you already produce: an OBS virtual camera, a second OBS output, a webcam over the table, or an RTSP, RTMP or SRT feed. It works alongside Whatnot, Fanatics Live, eBay Live, TikTok Shop, Twitch and YouTube Live without access to your accounts, and it is not affiliated with any of them. The browser interface shows each card as it settles and keeps a history of the show; an optional flag adds USD price statistics to every match. Identification is a paid API call, so it should not run on every frame. At 15 frames per second, a per-frame pipeline makes 54,000 calls an hour. Cardstream.ai finds the card with a local segmentation model, waits until it stops moving, and checks whether it is the same card as before. Only a new card triggers a call: about 100 an hour. Everything except that call runs on your CPU, no GPU needed, and only one image crop per new card leaves your machine. Identification uses Ximilar's Collectibles Recognition API out of the box; the identify step is one interface, so you can connect your own recogniser instead. Cardstream.ai itself is free, with code and model weights under Apache-2.0. It installs with one script on macOS and Linux, from PyPI on any platform, or as a Docker image. Shows can optionally be saved to your Ximilar account for review afterwards. Cardstream.ai is developed by Ximilar.
Listed in
Real-time card identification
Names each trading card on stream in about half a second: name, set, number, series and year
Supported games
25+ trading card games, including Pokémon, MTG, Yu-Gi-Oh!, One Piece and Lorcana
Confidence tiers
Every match is rated High, Medium or Low; set a threshold to keep weak matches off the overlay
Alternative matches
Up to four other cards it could have been, shown with every result
One API call per card
A local motion gate and same-card check mean only a new card triggers an identification call
Runs on CPU
Detection and gating run on your own computer, no GPU needed
Video sources
Webcam, video file, OBS push over RTMP or SRT, RTSP/RTMP/SRT pull, WebSocket JPEG feeds
Stream overlay
Browser UI with the match, a bounding box and a history of every card shown
Live settings
Thresholds, game and set code can be changed from the browser mid-stream
Price statistics
Optional USD median, range and latest sale for each match
Private by design
Only one image crop per new card leaves your machine; the stream stays local
Open source
Code and model weights under Apache-2.0; connect Ximilar’s API or your own recogniser
Cardstream.ai is built by Ximilar, a computer vision company in Brno, Czech Republic, and the maker of the Collectibles Recognition API. In live card breaks the seller opens packs on camera and names every card by hand. Cardstream.ai identifies those cards from the stream automatically. Most of the engineering went into keeping it cheap to run, so that the paid API is called only when a new card appears. It was first released as open source in September 2026.
Cardstream.ai is open source and runs on your own computer. It watches the video you already stream and identifies each trading card in about half a second, with no button to press. A local engine checks whether the card has stopped moving and whether it’s a new card. Only a new card triggers a paid identification call: about 100 calls an hour instead of 54,000.
Most card recognition tools for live streams are either closed paid services or browser extensions that identify one card per click. Cardstream.ai is free and open source (Apache-2.0), works automatically on the live video, and needs no GPU. It never logs in to your marketplace accounts. Only one image crop per new card leaves your machine, and you can connect your own recogniser instead of Ximilar’s API.
Leave this empty. There are no named public users yet, and a made-up list would be easy to disprove.
Card breakers and live sellers on Whatnot, Fanatics Live, eBay Live and TikTok Shop, and card shops streaming on Twitch or YouTube. It’s also for developers who want to add card recognition to their own streaming or scanning setup.
Python, ONNX Runtime, an RF-DETR segmentation model trained by Ximilar for finding the card, and an embedding model for the same-card check (Torch, ONNX or LiteRT backends). It also uses ffmpeg for RTSP, RTMP and SRT streams, a local browser interface, Docker, and Ximilar’s Collectibles Recognition API for identification.
We have collected here some useful links to help you find out if CardStream AI is good.
Check the traffic stats of CardStream AI on SimilarWeb. The key metrics to look for are: monthly visits, average visit duration, pages per visit, and traffic by country. Moreoever, check the traffic sources. For example "Direct" traffic is a good sign.
Check the "Domain Rating" of CardStream AI on Ahrefs. The domain rating is a measure of the strength of a website's backlink profile on a scale from 0 to 100. It shows the strength of CardStream AI's backlink profile compared to the other websites. In most cases a domain rating of 60+ is considered good and 70+ is considered very good.
Check the "Domain Authority" of CardStream AI on MOZ. A website's domain authority (DA) is a search engine ranking score that predicts how well a website will rank on search engine result pages (SERPs). It is based on a 100-point logarithmic scale, with higher scores corresponding to a greater likelihood of ranking. This is another useful metric to check if a website is good.
The latest comments about CardStream AI on Reddit. This can help you find out how popualr the product is and what people think about it.
Do you know an article comparing CardStream AI to other products?
Suggest a link to a post with product alternatives.
Is CardStream AI good? This is an informative page that will help you find out. Moreover, you can review and discuss CardStream AI here. The primary details have been verified within the last quarter. So they could be considered up to date. If you think we are missing something, please use the means on this page to comment or suggest changes. All reviews and comments are highly encouranged and appreciated as they help everyone in the community to make an informed choice. Please always be kind and objective when evaluating a product and sharing your opinion.
