3 ms·
Probably something like Apple Vision Framework or Amazon Textract or Google's Cloud Vision. Tesseract does well under ideal conditions, but the world is messy.
by criddell 2y ago
Probably something like Apple Vision Framework or Amazon Textract or Google's Cloud Vision.
Tesseract does well under ideal conditions, but the world is messy.
- cratermoon 2y agoI was thinking ABBYY FineReader, but those, too. Instead of using VLMs or any sort of generative AI, they're build on good old-fashioned feature extraction and nearest neighbor classifiers such as the k-nearest neighbors algorithm. It's possible to build a working prototype of this technique using basic ML algorithms.