Thanks to @matthen2, I got to test out my new candidate DeOldify model on a use case that I think will be quite appealing for many people: wedding photos! Thanks Matt!pic.twitter.com/OEscz5XhiW
Show this thread
-
Amazing work
@citnaj. Simpleton Question: I was wondering — would it make sense to have a domain-classifier at the head end of the process to decide which model to use? Eg “Oh this looks like a portrait, a wedding portrait, so let me use the wedding portrait trained colorizer.”
Well yes, you could. But you could also just make that all part of the model itself if you train it right, end to end, and it would work better (more nuanced treatment per photo). That's really what I'm aiming for here.
-
-
Yeah thanks I don’t know what I’m talking about here. It just strikes me that there are a lot of domains of photos, and domain specific trained models might yield more accurate results. Surprising that a “general” model yields better results; counterintuitive to me. Fascinating.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.