In my high school years, before I was a tech officer at Tri-City Band Corps, an outside of school marching band, I was a music librarian. The music librarian is someone who handles the music.
Whenever we, the group, bought new music, we would scan it and save it to our music collection. However, it was tedious to name the music because we had to follow a format, ie. SongName_Instrument_PartNumber.pdf. After scanning the music sheets, I wanted TesseractOCR (optical character recognition) to scan all the music, extract the necessary info (composer, song title, instrument, part number, etc.) and name the digital file appropriately.
I wrote the program with Java and it sort of worked. I had around a 70% success rate when I tried, but mostly the success rate depended on the clarity of the image.