Attention Consistency on Visual Corruptions for Single-Source Domain Generalization.

IEEE Conference on Computer Vision and Pattern Recognition(2022)

引用 24|浏览64
暂无评分
摘要
Generalizing visual recognition models trained on a single distribution to unseen input distributions (i.e. domains) requires making them robust to superfluous correlations in the training set. In this work, we achieve this goal by altering the training images to simulate new domains and imposing consistent visual attention across the different views of the same sample. We discover that the first objective can be simply and effectively met through visual corruptions. Specifically, we alter the content of the training images using the nineteen corruptions of the ImageNet-C benchmark and three additional transformations based on Fourier transform. Since these corruptions preserve object locations, we propose an attention consistency loss to ensure that class activation maps across original and corrupted versions of the same training sample are aligned. We name our model Attention Consistency on Visual Corruptions (ACVC). We show that ACVC consistently achieves the state of the art on three single-source domain generalization benchmarks, PACS, COCO, and the large-scale DomainNet.
更多
查看译文
关键词
visual corruptions,domain
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要