I’m really drawn to the text that draws the two lines of thought in photographic archiving strategies. Firstly, I had only recently become aware of MCAD when discovering my program director is an alum, so its interesting to see it appear again. It also makes me think about how much Midwest design and art (and archiving!) is underrepresented against the larger coastal institutions, whose money is old and vast.
I’m also drawn to the idea of archiving images by their visual semantics. As a technologist, I always think about how machines “see” and process information: They are highly semantic! At their core, computers see one small block of color in a long single line of colors. With some math, they see edges, they see gradients:
The goal of computer vision, CV, is to get context. The largest competition/effort in CV is COCO , or Common Object in Context, it’s right there in the name! So, at least for machine learning, the visual semantic is the base and necessary component for contextual organization. I guess I’m curious if machines could implement the discourse of the document, or is that soley a human task? Could the NYPL scan their magazine clippings and be told what heading it is, especially when they have wild headings like Views from Behind?
Thanks, Iltimas! I’m glad the Kamin reading resonated for you. I wonder, though: do machines see semantically? Are they looking for *meaning*, or for form — e.g., edges, gradients, etc.? And I love this idea of training an AI on the NYPL Pictures Collection heading and seeing what it learns; I imagine it would generate some kind of hilarious Janelle Shane-style AI Weirdness!
And this is your fourth and final post — thank you!