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

[1]

LI H., Analysis of image recognition technology in artificial intelligence, Wireless Internet Technology, 18, 17, pp. 93-94, (2021)

[2]

REN J M, HU X L., Application of image recognition technology in the field of public safety, Computer Products and Distribution, 5, (2019)

[3]

WEI X S, LUO J H, WU J X, Et al., Selective convolutional descriptor aggregation for fine-grained image retrieval, IEEE Transactions on Image Processing:A Publication of the IEEE Signal Processing Society, 26, 6, pp. 2868-2881, (2017)

[4]

JI X, VEDALDI A, HENRIQUES J., Invariant information clustering for unsupervised image classification and segmentation, Proceedings of the IEEE/CVF International Conference on Computer Vision, pp. 9864-9873, (2019)

[5]

LIN T Y, ROYCHOWDHURY A, MAJI S., Bilinear CNN models for fine-grained visual recognition, Proceedings of the IEEE International Conference on Computer Vision, pp. 1449-1457, (2015)

[6]

LUO J H, WU J X., A survey on fine-grained image categorization using deep convolutional features, Acta Automatica Sinica, 43, 8, pp. 1306-1318, (2017)

[7]

MAJI S, RAHTU E, KANNALA J, Et al., Fine-grained visual classification of aircraft, Proceedings of the IEEE International Conference on Computer Vision, pp. 1-6, (2013)

[8]

HAN B, YAO Q, YU X, Et al., Co-teaching: Robust training of deep neural networks with extremely noisy labels, Advances in Neural Information Processing Systems, 3, pp. 8527-8537, (2018)

[9]

ZHANG Y, WEI X S, WU J X, Et al., Weakly supervised fine-grained categorization with part-based image representation, IEEE Transactions on Image Processing:A Publication of the IEEE Signal Processing Society, 25, 4, pp. 1713-1725, (2016)

[10]

SIMON M, RODNER E., Neural activation constellations: Unsupervised part model discovery with convolutional networks, Proceedings of the IEEE International Conference on Computer Vision, pp. 1143-1151, (2015)

← 1 2 3 4 →