πŸ—‘οΈ Delete Fingerprints?

This will remove all hashes for this game. You can rebuild anytime.

🧠 Card Recognition Database

Build a local database of card image fingerprints for high-accuracy recognition (95%+). Downloads thumbnails from 7 free TCG APIs β€” no paid accounts needed. Supports PokΓ©mon, MTG, Yu-Gi-Oh!, Flesh & Blood, Lorcana, One Piece & Dragon Ball Super.

88600
Cards Fingerprinted
1402
Sets Covered
20185
PokΓ©mon Cards
58966
MTG Cards
9206
Yu-Gi-Oh! Cards
~68,000
Total Available

Build Fingerprint Database

βœ… Database ready! 88600 cards fingerprinted across 1402 sets.
βœ“ Done
🎴
PokΓ©mon TCG
20185 cards
βœ“ Done
πŸ§™
Magic: The Gathering
58966 cards
βœ“ Done
πŸ‘οΈ
Yu-Gi-Oh!
9206 cards
Empty
βš”οΈ
Flesh & Blood
0 cards
βœ“ Done
✨
Lorcana
243 cards
Empty
πŸ΄β€β˜ οΈ
One Piece TCG
0 cards
Empty
πŸ‰
Dragon Ball Super
0 cards
Click a game card above to select, or build all at once

How Image Fingerprinting Works

πŸ“₯
Step 1: Download

Fetches thumbnails from 7 free TCG APIs β€” PokΓ©mon, MTG, YGO, FaB, Lorcana, One Piece, DBS

πŸ”’
Step 2: Fingerprint

Computes a 256-bit perceptual hash for each image β€” unique per card across all games

πŸ“Έ
Step 3: Scan

When you scan a card, we compute its hash and compare against ~68,000 fingerprints

🎯
Step 4: Match

Closest hash distance = card identity. Works on holographic cards where OCR fails

Recognition Priority

1st β€” 95%+
Hash Fingerprint Match (requires this database to be built)
2nd β€” 80%+
Card Number OCR Extract "086/198" from card bottom, exact API lookup
3rd β€” 50%+
Name OCR Read card name with heavy preprocessing, fuzzy API match
Fallback
Manual Review Card saved for human review β€” garbage text is NEVER saved