Neural Inverse Knitting: From Images to Manufacturing Instructions

Alexandre Kaspar, Tae-Hyun Oh, Liane Makatura, Petr Kellnhofer, Wojciech Matusik
Proceedings of the 36th International Conference on Machine Learning, PMLR 97:3272-3281, 2019.

Abstract

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

Cite this Paper

BibTeX
@InProceedings{pmlr-v97-kaspar19a, title = {Neural Inverse Knitting: From Images to Manufacturing Instructions}, author = {Kaspar, Alexandre and Oh, Tae-Hyun and Makatura, Liane and Kellnhofer, Petr and Matusik, Wojciech}, booktitle = {Proceedings of the 36th International Conference on Machine Learning}, pages = {3272--3281}, year = {2019}, editor = {Chaudhuri, Kamalika and Salakhutdinov, Ruslan}, volume = {97}, series = {Proceedings of Machine Learning Research}, month = {09--15 Jun}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v97/kaspar19a/kaspar19a.pdf}, url = {https://proceedings.mlr.press/v97/kaspar19a.html}, abstract = {Motivated by the recent potential of mass customization brought by whole-garment knitting machines, we introduce the new problem of automatic machine instruction generation using a single image of the desired physical product, which we apply to machine knitting. We propose to tackle this problem by directly learning to synthesize regular machine instructions from real images. We create a cured dataset of real samples with their instruction counterpart and propose to use synthetic images to augment it in a novel way. We theoretically motivate our data mixing framework and show empirical results suggesting that making real images look more synthetic is beneficial in our problem setup.} }
Endnote
%0 Conference Paper %T Neural Inverse Knitting: From Images to Manufacturing Instructions %A Alexandre Kaspar %A Tae-Hyun Oh %A Liane Makatura %A Petr Kellnhofer %A Wojciech Matusik %B Proceedings of the 36th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2019 %E Kamalika Chaudhuri %E Ruslan Salakhutdinov %F pmlr-v97-kaspar19a %I PMLR %P 3272--3281 %U https://proceedings.mlr.press/v97/kaspar19a.html %V 97 %X Motivated by the recent potential of mass customization brought by whole-garment knitting machines, we introduce the new problem of automatic machine instruction generation using a single image of the desired physical product, which we apply to machine knitting. We propose to tackle this problem by directly learning to synthesize regular machine instructions from real images. We create a cured dataset of real samples with their instruction counterpart and propose to use synthetic images to augment it in a novel way. We theoretically motivate our data mixing framework and show empirical results suggesting that making real images look more synthetic is beneficial in our problem setup.
APA
Kaspar, A., Oh, T., Makatura, L., Kellnhofer, P. & Matusik, W.. (2019). Neural Inverse Knitting: From Images to Manufacturing Instructions. Proceedings of the 36th International Conference on Machine Learning, in Proceedings of Machine Learning Research 97:3272-3281 Available from https://proceedings.mlr.press/v97/kaspar19a.html.

Related Material