Mixup linearly interpolates pairs of examples to form new samples . It can achieve a strong baseline with less trainingtime than original mixup . mWh can also transfer to CutMix,and gain consistent improvement on other machine learning and computer vision tasks such as object detection . Our code is open-source and available athttps://://://www.gong.com/yuhao318/mwh.mwh and the code is available athttp://://gongaoao.org//mWh . We show that mWh strikes a goodbalance between exploration and exploitation by gradually replacing mixup withbasic data augmentation . It is also possible to use CutMix as a training algorithm for other machine-learning tasks like object-detection and object-recognition tasks .

Author(s) : Hao Yu, Huanyu Wang, Jianxin Wu

Links : PDF - Abstract

Code :
Coursera

Keywords : mwh - mixup - object - tasks - detection -

Leave a Reply

Your email address will not be published. Required fields are marked *