Long-Tailed Visual Recognition via Improved Cross-Window Self-Attention and TrivialAugment

doi:10.1109/ACCESS.2023.3277204

	Long-Tailed Visual Recognition via Improved Cross-Window Self-Attention and TrivialAugment
	Song, Ying 1,2,3; Li, Mengxing 1,2; Wang, Bo 4
	2023
发表期刊	IEEE ACCESS
ISSN	2169-3536
卷号	11 页码:49601-49610
摘要	In the real world, large-scale image data sets usually present long-tailed distribution. When traditional visual recognition methods are applied to long-tail image data sets, problems such as model failure and sudden decline in recognition accuracy occur. While, when deep learning models encounter long-tailed datasets, they tend to perform poorly. In order to mitigate the impact of these problems, we propose CWTA (Long-tailed Visual Recognition via improved Cross-Window Self-Attention and TrivialAugment). CWTA uses CNN to better capture the local features of the image, uses the Cross-Window Self-Attention mechanism to dynamically adjust the perception domain to better deal with image noise, and uses TrivialAugment to enhance the diversity of a few types of data samples, thus improving the recognition accuracy of long-tailed distributed images. The experimental results show that the proposed CWTA performs best in the classification accuracy of different categories on different long-tailed datasets. We also compared CWTA with other long-tailed recognition algorithms (such as OLTR, LWS, ResLT, PaCo, and BALLAD), and the CWTA is the best when ResNet-50 as the Backbone. On the CIFAR100-LT, ImageNet-LT, and Places-LT datasets, the acc of all categories of CWTA is 12.9%, 0.4%, and 1.3% higher than that of BALLAD, respectively. For F-1-Score on CIFAR100-LT, ImageNet-LT, and Places-LT datasets, CWTA is 6.6%, 2.2%, and 1.5% higher than BALLAD, respectively.
关键词	Convolutional neural networks Continuous wavelet transforms Data models Training Computer vision Transformers Transfer learning Long-tailed recognition self-attention vision transformer CNN TrivialAugment
DOI	10.1109/ACCESS.2023.3277204
收录类别	SCI
语种	英语
资助项目	National Natural Science Foundation of China[61872043] ; State Key Laboratory of Computer Architecture, Institute of Computing Technology (ICT), Chinese Academy of Sciences (CAS)[CARCHA202103]
WOS研究方向	Computer Science ; Engineering ; Telecommunications
WOS类目	Computer Science, Information Systems ; Engineering, Electrical & Electronic ; Telecommunications
WOS记录号	WOS:001005708800001
出版者	IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
引用统计
文献类型	期刊论文
条目标识符	http://119.78.100.204/handle/2XEOYT63/21194
专题	中国科学院计算技术研究所期刊论文_英文
通讯作者	Song, Ying
作者单位	1.Beijing Informat Sci & Technol Univ, Beijing Key Lab Internet Culture & Digital Dissemi, Beijing 100101, Peoples R China 2.Beijing Informat Sci & Technol Univ, Beijing Adv Innovat Ctr Mat Genome Engn, Beijing 100101, Peoples R China 3.Chinese Acad Sci, Inst Comp Technol, State Key Lab Comp Architecture, Beijing 100086, Peoples R China 4.Zhengzhou Univ Light Ind, Software Engn Coll, Zhengzhou 450002, Peoples R China
推荐引用方式 GB/T 7714	Song, Ying,Li, Mengxing,Wang, Bo. Long-Tailed Visual Recognition via Improved Cross-Window Self-Attention and TrivialAugment[J]. IEEE ACCESS,2023,11:49601-49610.
APA	Song, Ying,Li, Mengxing,&Wang, Bo.(2023).Long-Tailed Visual Recognition via Improved Cross-Window Self-Attention and TrivialAugment.IEEE ACCESS,11,49601-49610.
MLA	Song, Ying,et al."Long-Tailed Visual Recognition via Improved Cross-Window Self-Attention and TrivialAugment".IEEE ACCESS 11(2023):49601-49610.