Abstract

Motivated by the recent potential of mass customization brought bywhole-garment knitting machines, we introduce the new problem of automaticmachine instruction generation using a single image of the desired physicalproduct, which we apply to machine knitting. We propose to tackle this problemby directly learning to synthesize regular machine instructions from realimages. We create a cured dataset of real samples with their instructioncounterpart and propose to use synthetic images to augment it in a novel way.We theoretically motivate our data mixing framework and show empirical resultssuggesting that making real images look more synthetic is beneficial in ourproblem setup.

 

Quick Read (beta)