smoke_signal/tessdata/README.md

Smoke Signal Tessdata Drop Folder

Put custom Tesseract model files here as .traineddata.

Naming rule:

• Use lowercase + underscores based on detected font name.
• Example: Gill Sans Infant -> gill_sans_infant.traineddata

The OCR runtime will auto-pick models from either:

• /tmp/smoke_signal/tessdata/
• smoke_signal/tessdata/ (this folder)

Important:

• .otf / .ttf files are not Tesseract models.
• A file like gill_sans_infant.traineddata.otf is still a font file, not usable by Tesseract OCR.

If your model name is custom and does not match slug, register it once:

from smoke_signal.scripts.font_library import register_font_model
register_font_model("Gill Sans Infant", model_name="my_model_name")