I know about Google goggles, but I need something I can train using lesser
known features (i.e. local stuff that's important to us but that others
don't really know about) to help with processing of archival photographs.
Does such software exist in a form that would be accessible to libraries?

Thanks,

kyle