What is font training?


When you buy Scan2CAD, its OCR text recognition is able to recognize a range of standard fonts. If your raster images contain fonts that Scan2CAD cannot recognize or cannot recognize well, you can train Scan2CAD to recognize them.


Neural networks

Scan2CAD recognizes text using technology called neural networks. Neural networks work in an analogous way to the human brain. They learn to recognize shapes and patterns from a range of examples.

Font training allows you to train a neural network to recognize a font from examples of that font. You can train as many neural networks as you like to recognize as many fonts as you like.


One neural network or many?

While you can train a single neural network to recognize a variety of fonts, text recognition is likely to be far more successful if you train different neural networks to recognize different fonts, unless the fonts are very similar.

For optimal results you will probably also need to train different neural networks to cater for extreme differences in character size and normal and italic versions of the same font.


Handwritten text

You can train a neural network to recognize handwritten or stenciled text so long as it is not “joined up”.

Where text has been written freehand it is likely to be consistent to the individual who wrote it. Where a range of drawings has been produced by an individual you can train a neural network to recognize that individual’s writing.

Where text has been stenciled, you can train one or several neural networks to recognize it depending on the similarity between different stencil types.


What font training can’t do

You cannot train a neural network to recognize “joined up” writing or poor quality text. It can only be trained to recognize distinct and complete characters that do not touch, run or bleed into each other.


Have questions on this topic? Talk to us