Plate Configuration File¶

The plate config file defines how license plate images and text should be preprocessed for OCR model training and inference.

This file is parsed using the PlateOCRConfig class and validated with Pydantic.

Config Fields¶

Below is a summary of the supported fields in the YAML config:

Field	Type	Description
`max_plate_slots`	`int`	Maximum number of characters the model can recognize on a plate
`alphabet`	`str`	The full set of characters that the model can output (no duplicates)
`pad_char`	`str`	A single character used for padding shorter plate texts
`img_height`	`int`	Height of input images
`img_width`	`int`	Width of input images
`keep_aspect_ratio`	`bool`	Whether to keep the original aspect ratio of the image
`interpolation`	`"cubic"`, `"linear"`, etc.	Resizing interpolation method
`image_color_mode`	`"grayscale"` or `"rgb"`	Color mode for input images
`padding_color`	`int` or `(int, int, int)`	Color used to pad the image if aspect ratio is preserved

Config Example¶

plate_config.yaml

max_plate_slots: 9
alphabet: "0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ_"  # (1)!
pad_char: "_"  # (2)!
img_height: 64
img_width: 160
keep_aspect_ratio: true
interpolation: linear  # (3)!
image_color_mode: grayscale
padding_color: 114  # (4)!

All the possible character set for the model output. Must include the pad character.
Padding character for plates which length is smaller than MAX_PLATE_SLOTS.
Matches OpenCV's interpolation names.
Only used when keep_aspect_ratio is True.

Tip

For examples used in the default models, checkout config directory (located in root dir of project).