We’ve been doing extensive work on manipulating glyph images in a variety of ways in recent weeks. We’re doing this so we can produce a large library of samples of each individual glyph to train the AI (artificial intelligence) and machine learning code we’ll be using later on down the road for glyph detection. For any single glyph, we’re currently creating between 60,000 and 240,000 image variations. Multiply that by 1,000 glyphs and you get up to a total of 240 million images that will be created. For the techies out there, the data storage requirements for this are fairly substantial. We estimate data storage requirements to be in the 2TB to 3TB range for images once they’re all generated. The computational requirements for processing the image libraries will also be substantial. We should have more information on that in the next month or so.
Glyph Image Manipulation
Share This
