John Resig’s post “Using Computer Vision to Increase the Research Potential of Photo Archives” was both inspiring and eye-opening. Among other things, it triggered the realization that there may already exist digital tools that would simplify some of the more painstaking and error-prone tasks that my work on the Harry Callahan survey comprises. One of the overarching complications of this project is the need to aggregate a set of collection records from ten separate institutions whose own styles and standards—of cataloging, of digitizing—are widely divergent. Another is the task of linking multiple copies of the same image across collections, despite potential discrepancies in metadata. Much of this work I’ve been resigned to doing by hand and eye, but, again, I’m beginning to realize that there may be a better way. Call it a first step into the light of Digital Art History? Tools to corral and correct the data even before the work of interpretation and dissemination begins.

As a test case for the Building a Digital Portfolio institute, I’m curious to see what I can do with these piles of placeholder thumbnail images that I’ve culled from the checklists of the collections included in our survey. They’re pretty raw as data goes, or clumsy at least, currently severed from their associated metadata, and all over the map in terms of image quality. What can I make of them in the next two weeks? What can they suggest about the potential for this project, and the tools that might speed it on its way?

Source: a project wish list