CutMix
0 订阅
CutMix is an image data augmentation strategy. Instead of simply removing pixels as in Cutout, we replace the removed regions with a patch from another image. The ground truth labels are also mixed proportionally to the number of pixels of combined images. The added patches further enhance localization ability by requiring the model to identify the object from a partial view.
相关学科: MixupSemi-Supervised Semantic SegmentationAutoAugmentCascade R-CNNCutoutAdvPropFMixMishDropBlockFine-Grained Image Recognition
学科讨论

暂无讨论内容,你可以
推荐文献
按被引用数
学科管理组
暂无学科课代表,你可以申请成为课代表
重要学者
Alan L. Yuille
56169 被引用,870
篇论文
Bernt Schiele
53686 被引用,602
篇论文
Michael Jones
35461 被引用,132
篇论文
Jianfeng Gao
30435 被引用,454
篇论文
Chen Chen
23149 被引用,969
篇论文
Sanjay Singh
19371 被引用,1361
篇论文
Yun Fu
18591 被引用,525
篇论文
Lei Zhang
13050 被引用,317
篇论文
Wenyu Liu
12725 被引用,365
篇论文
Amit Singh
12259 被引用,761
篇论文