3 ms·
Impressive stuff, I especially like the style transfer that can be done by using the global features of one image and the local features of another (Fig. 7) Wh
by astrosi 10y ago
Impressive stuff, I especially like the style transfer that can be done by using the global features of one image and the local features of another (Fig. 7)
What I find somewhat annoying is that whilst they show some examples from their validation set, and a couple of examples of the model failures. They don't appear to show a random selection of cases from their validation set.
- argonaut 10y agoThe results that I see seem quite good, but I also find the lack of any objective benchmarks or performance measures to be one flaw in the paper. This might explain why they published in a graphics conference (SIGGRAPH) rather than an ML or computer vision conference (which would probably demand benchmarks).
- babo 10y agoWhy not trying it out yourself? They provide the model to download and torch code to run it on any photo. https://github.com/satoshiiizuka/siggraph2016_colorization https://github.com/satoshiiizuka/siggraph2016_colorization