CardSight AI adds language detection for Pokémon, One Piece and Magic cards

CardSight AI has added printed-language detection to card identification across Pokémon, the One Piece Card Game and Magic: The Gathering. The update is meant to stop apps from silently mislabeling Japanese and other non-English cards as U.S. versions.

Why it matters: - Card scanners and marketplace tools can now distinguish cards that share artwork but differ by printed language. - The change reduces silent misidentification that can affect collection tracking, pricing and listings. - Developers can now support multilingual card identification without building separate regional products.

What happened: - CardSight AI announced language detection for card identification across Pokémon, the One Piece Card Game and Magic: The Gathering. - Identification results now report the printed language of scanned cards across eight supported languages: English, French, German, Italian, Spanish, Japanese, Korean and Chinese. - The update is live across CardSight AI identification endpoints.

The details: - A scan of a Japanese card now returns the U.S. canonical card plus the detected Japanese print. - Pokémon language detection covers every published language CardSight AI supports, including Japanese, Korean and Chinese. - One Piece Card Game language detection covers Japanese and every other supported language for that game. - Magic: The Gathering language detection covers all eight supported languages. - The new capability uses the same identification endpoints developers already use. - Existing integrations do not need a new setup or per-region product maintenance. - CardSight AI said official Japanese identification, where those sets carry their own catalog records, follows this release. - The company said it shipped identification support for Pokémon’s Pitch Black set within days of the set reaching shelves worldwide on July 17. - CardSight AI said it trains its AI specifically on trading cards, which lets new coverage extend existing systems. - CardSight AI says identification returns in under 300 milliseconds. - Existing solutions commonly take several seconds per card, according to the company. - The platform’s “Break Out of the Box” campaign supports angled photos, multiple cards in one image and full nine-card binder pages in a single call.

Between the lines: - The release targets a common weak point in visual-only card recognition, where lookalike cards can collapse into the same result. - Language detection gives marketplaces and collectors a clearer signal on variants that look identical but differ in catalog identity and value. - The update also strengthens CardSight AI’s pitch to developers that need fast global coverage from one API instead of multiple regional tools. - Eric Nusbaum, co-founder and CEO, said a system that returns a U.S. record for a Japanese card without warning is guessing, not identifying. - Signe Bone, founding engineer, said the company designed the system so new languages extend the platform instead of requiring a rewrite.

What’s next: - CardSight AI says more exact Japanese identification will follow for sets that have their own catalog records. - Developers can use the free tier with 750 API calls and no credit card. - The company also offers transparent usage pricing, REST access, SDKs for Node.js, Python, Swift, Java and .NET, plus MCP endpoints. - Pokémon, One Piece and Magic now sit alongside CardSight AI’s sports card API and shared trading card database for teams serving collectors worldwide.

The bottom line: - CardSight AI is trying to make multilingual card scanning accurate enough for real collection, pricing and marketplace workflows, not just close enough for image matching. - The company’s bet is that language-aware identification becomes table stakes as trading card hobby tools go global.