Fine-grained image classification method with noisy labels based on retrieval augmentation

被引：0

作者：

Bao, Heng ^{[1
]}

Deng, Lirui ^{[2
]}

Zhang, Liang ^{[3
]}

Chen, Xunxun ^{[3
]}

机构：

[1] School of Cyber Security, University of Chinese Academy of Sciences, Beijing

[2] Department of Computer Science and Technology, Tsinghua University, Beijing

[3] National Computer Network Emergency Response Technical Team, Coordination Center of China, Beijing

来源：

Beijing Hangkong Hangtian Daxue Xuebao/Journal of Beijing University of Aeronautics and Astronautics | 2024年 / 50卷 / 07期

关键词：

data cleaning; fine-grained image classification; image retrieval; network security; noisy labels;

D O I：

10.13700/j.bh.1001-5965.2022.0589

中图分类号：

学科分类号：

摘要：

In the application of Internet audio and video content analysis, it is of great significance to establish a fast fine-grained image classification method with lowlabeling costs. Due to the more similar appearance features between categories and the existence of interference factors such as illumination, viewing angle, and background occlusion, fine-grained image classification faces challenges such as large number of categories, small differences between categories, high labeling cost, and low label signal-to-noise ratio. In order to improve the effect of fine-grained classification of massive images in a data environment with noisy labels, a fine-grained image classification method based on retrieval augmentation was proposed. Based on iterative cleaning of noisylabels, the retrieval paradigm was used to obtain more expressive features through simple category labeling, so as to improve the recognition ability of the classifier. In addition, favorableresults wereachieved on the dataset containing 1 500 fine-grained food categories and more than 500 000 images. © 2024 Beijing University of Aeronautics and Astronautics (BUAA). All rights reserved.

引用

页码：2284 / 2292

页数：8

共 38 条

[21]

BRANSON S, VAN HORN G, BELONGIE S, Et al., Bird species categorization using pose normalized deep convolutional nets

[22]

ZHUANG P Q, WANG Y L, QIAO Y., Learning attentive pairwise interaction for fine-grained classification, AAAI Conference on Artificial Intelligence, 34, 7, pp. 13130-13137, (2020)

[23]

BAI Y Y, LIU N Z, JIANG X T., Fine grained image classification network combined with attention CutMix, Computer Technology and Development, 31, 10, pp. 38-42, (2021)

[24]

ZHU H W, KE W J, LI D, Et al., Dual cross-attention learning for fine-grained visual categorization and object re-identification, Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 4682-4692, (2022)

[25]

DING Y, ZHOU Y Z, ZHU Y, Et al., Selective sparse sampling for fine-grained image recognition, Proceedings of the IEEE/CVF International Conference on Computer Vision, pp. 6598-6607, (2019)

[26]

CHANG D L, DING Y F, XIE J Y, Et al., The devil is in the channels: Mutual-channel loss for fine-grained image classification, IEEE Transactions on Image Processing: A Publication of the IEEE Signal Processing Society, pp. 4683-4695, (2020)

[27]

LI H F., Research on reordering algorithms in image retrieval, pp. 7-17, (2015)

[28]

JING Y S, BALUJA S., VisualRank: Applying PageRank to large-scale image search, IEEE Transactions on Pattern Analysis and Machine Intelligence, 30, 11, pp. 1877-1890, (2008)

[29]

AMBAI M, YOSHIDA Y., Multiclass VisualRank: Image ranking method in clustered subsets based on visual features, Proceedings of the 32nd International ACM SIGIR Conference on Research and Development in Information Retrieval, pp. 732-733, (2009)

[30]

CHUM O, PHILBIN J, SIVIC J, Et al., Total recall: Automatic query expansion with a generative feature model for object retrieval, Proceedings of the IEEE 11th International Conference on Computer Vision, pp. 1-8, (2007)

← 1 2 3 4 →