Hi Peter,
Be sure to look at the resources and tools that are discussed through the IS&T (Society for Imaging Science and Technology) (imaging.org) and esp. research and tools developed at RIT (Rochester Institute of Technology) where many of the issues that you're thinking about are dealt with regularly. 

Kari Smith
MIT Institute Archives and Special Collections 

-----Original Message-----
From: Code for Libraries [mailto:[log in to unmask]] On Behalf Of Peter Mangiafico
Sent: Wednesday, July 22, 2015 6:50 PM
To: [log in to unmask]
Subject: [CODE4LIB] survey of image analysis packages

I am conducting a survey of software used for image analysis and metadata enhancement.  Examples include facial recognition, object identification, similarity matching, and so on.  The goal is to understand if it is possible to use algorithmic techniques to improve discoverability in a large dataset consisting mostly of images.  The main project I am working on is automobile history (http://revslib.stanford.edu <http://revslib.stanford.edu/> <http://revslib.stanford.edu/ <http://revslib.stanford.edu/>>) but the techniques can of course be applied much more widely.  I'm interested in a broad sweep, could be open source, commercial, service model, API, etc.  If you have projects you are aware of, or tools you have used or heard about, and wouldn't mind sending me an email, I'd appreciate it!  Thanks,

Peter Mangiafico
Stanford University Libraries
[log in to unmask] <mailto:[log in to unmask]>