Ask HN: How is Apple's text recognition so good?
My iPhone's ability to quickly identify text across the thousands of photos I have taken is very good. For example, today I searched a word that was found in small print in a newspaper I had taken a photo of. What makes this technology so good? Is it possible to build something similarly good myself? Are there any great open-source text recognition packages? Apple sometimes relies on mainframe computers to supplement the work of its personal devices. The this sort of communication and cooperation can happen without your being aware. I suspect that the results of this sort of work can accumulate over considerable time allowing Apple's devices to appear more powerful than they actually are. Perhaps this is the answer. I don't know but I do know that LLMs like chatgpt can be used to correct obvious errors in the way a human could, for example splitting words that should have a space between but don't, random wrong char4cters etc.