Half-assed datasets continue to be a large part of the problems in ML.
When doing colorization algorithm comparisons, just looking at results on images from modern datasets like Imagenet seems to miss the point... I'll continue doing me but it's weird to see this happen over and over again.
-
-
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
@citnaj here's an idea that a friend suggested: why not try and work in the YUV color space and by edge detection, you could create Hue and Saturation templates that you could then flood fill with levels of your choice -
There's an idea I have that's similar in spirit- in trying to achieve user directed colorization- that I plan on exploring within the next few months. It's something I've been thinking about for a while but time/resources as usual are the limiting factor.
- Show replies
New conversation -
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.