Collected sources and patterns will appear here. Add from search, explore, or the patterns library.
Image -> Image
Synthesize localized anomalous features on normal images by optimizing pixel alterations via gradient ascent against a target loss.
Problem it solves
Deep anomaly models require synthetic training data when real physical anomalies are unavailable.
Consumes
Emits
The real projects this mechanism was found in. Attribution is the point — this is how the best teams actually do it.