feat: improved local OCR with Strip & Match distillery detection

Browse Source

- Added comprehensive distillery database (200+ entries)
- Implemented Strip & Match heuristic for fuzzy matching
- Added contextual age detection from distillery lines
- Added whitespace normalization for OCR text
- Disabled local name extraction (too noisy, let Gemini handle it)
- Fixed confidence scale normalization in TastingEditor (0-1 vs 0-100)
- Improved extractName filter (60% letters required)
- Relaxed Fuse.js thresholds for partial matches

...

This commit is contained in:

robin

2025-12-25 13:14:08 +01:00

parent a1a91795d1

commit afe9197776

17 changed files with 3642 additions and 262 deletions

Show all changes Ignore whitespace when comparing lines Ignore changes in amount of whitespace Ignore changes in whitespace at EOL

Download Patch File Download Diff File Expand all files Collapse all files

BIN

public/tessdata/eng.traineddata Normal file

Binary file not shown.

BIN

public/tessdata/eng.traineddata.gz Normal file

Binary file not shown.

191

public/tessdata/tesseract-core-lstm.wasm.js Normal file

File diff suppressed because one or more lines are too long

191

public/tessdata/tesseract-core-relaxedsimd-lstm.wasm.js Normal file

File diff suppressed because one or more lines are too long

191

public/tessdata/tesseract-core-relaxedsimd.wasm.js Normal file

File diff suppressed because one or more lines are too long

191

public/tessdata/tesseract-core-simd-lstm.wasm.js Normal file

File diff suppressed because one or more lines are too long

191

public/tessdata/tesseract-core-simd.wasm.js Normal file

File diff suppressed because one or more lines are too long

191

public/tessdata/tesseract-core.wasm.js Normal file

File diff suppressed because one or more lines are too long