You Only Cut Once: Boosting Data Augmentation with a Single Cut

Junlin Han, Pengfei Fang, Weihao Li, Jie Hong, Mohammad Ali Armin, Ian Reid, Lars Petersson, Hongdong Li
Proceedings of the 39th International Conference on Machine Learning, PMLR 162:8196-8212, 2022.

Abstract

We present You Only Cut Once (YOCO) for performing data augmentations. YOCO cuts one image into two pieces and performs data augmentations individually within each piece. Applying YOCO improves the diversity of the augmentation per sample and encourages neural networks to recognize objects from partial information. YOCO enjoys the properties of parameter-free, easy usage, and boosting almost all augmentations for free. Thorough experiments are conducted to evaluate its effectiveness. We first demonstrate that YOCO can be seamlessly applied to varying data augmentations, neural network architectures, and brings performance gains on CIFAR and ImageNet classification tasks, sometimes surpassing conventional image-level augmentation by large margins. Moreover, we show YOCO benefits contrastive pre-training toward a more powerful representation that can be better transferred to multiple downstream tasks. Finally, we study a number of variants of YOCO and empirically analyze the performance for respective settings.

Cite this Paper

BibTeX
@InProceedings{pmlr-v162-han22a, title = {You Only Cut Once: Boosting Data Augmentation with a Single Cut}, author = {Han, Junlin and Fang, Pengfei and Li, Weihao and Hong, Jie and Armin, Mohammad Ali and Reid, Ian and Petersson, Lars and Li, Hongdong}, booktitle = {Proceedings of the 39th International Conference on Machine Learning}, pages = {8196--8212}, year = {2022}, editor = {Chaudhuri, Kamalika and Jegelka, Stefanie and Song, Le and Szepesvari, Csaba and Niu, Gang and Sabato, Sivan}, volume = {162}, series = {Proceedings of Machine Learning Research}, month = {17--23 Jul}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v162/han22a/han22a.pdf}, url = {https://proceedings.mlr.press/v162/han22a.html}, abstract = {We present You Only Cut Once (YOCO) for performing data augmentations. YOCO cuts one image into two pieces and performs data augmentations individually within each piece. Applying YOCO improves the diversity of the augmentation per sample and encourages neural networks to recognize objects from partial information. YOCO enjoys the properties of parameter-free, easy usage, and boosting almost all augmentations for free. Thorough experiments are conducted to evaluate its effectiveness. We first demonstrate that YOCO can be seamlessly applied to varying data augmentations, neural network architectures, and brings performance gains on CIFAR and ImageNet classification tasks, sometimes surpassing conventional image-level augmentation by large margins. Moreover, we show YOCO benefits contrastive pre-training toward a more powerful representation that can be better transferred to multiple downstream tasks. Finally, we study a number of variants of YOCO and empirically analyze the performance for respective settings.} }
Endnote
%0 Conference Paper %T You Only Cut Once: Boosting Data Augmentation with a Single Cut %A Junlin Han %A Pengfei Fang %A Weihao Li %A Jie Hong %A Mohammad Ali Armin %A Ian Reid %A Lars Petersson %A Hongdong Li %B Proceedings of the 39th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2022 %E Kamalika Chaudhuri %E Stefanie Jegelka %E Le Song %E Csaba Szepesvari %E Gang Niu %E Sivan Sabato %F pmlr-v162-han22a %I PMLR %P 8196--8212 %U https://proceedings.mlr.press/v162/han22a.html %V 162 %X We present You Only Cut Once (YOCO) for performing data augmentations. YOCO cuts one image into two pieces and performs data augmentations individually within each piece. Applying YOCO improves the diversity of the augmentation per sample and encourages neural networks to recognize objects from partial information. YOCO enjoys the properties of parameter-free, easy usage, and boosting almost all augmentations for free. Thorough experiments are conducted to evaluate its effectiveness. We first demonstrate that YOCO can be seamlessly applied to varying data augmentations, neural network architectures, and brings performance gains on CIFAR and ImageNet classification tasks, sometimes surpassing conventional image-level augmentation by large margins. Moreover, we show YOCO benefits contrastive pre-training toward a more powerful representation that can be better transferred to multiple downstream tasks. Finally, we study a number of variants of YOCO and empirically analyze the performance for respective settings.
APA
Han, J., Fang, P., Li, W., Hong, J., Armin, M.A., Reid, I., Petersson, L. & Li, H.. (2022). You Only Cut Once: Boosting Data Augmentation with a Single Cut. Proceedings of the 39th International Conference on Machine Learning, in Proceedings of Machine Learning Research 162:8196-8212 Available from https://proceedings.mlr.press/v162/han22a.html.

Related Material