Feature Fusion from Head to Tail for Long-Tailed Visual Recognition

被引:0
|
作者
Li, Mengke [1 ,2 ]
Hu, Zhikai [3 ]
Lu, Yang [4 ]
Lan, Weichao [3 ]
Cheung, Yiu-ming [3 ]
Huang, Hui [2 ]
机构
[1] Guangdong Lab Artificial Intelligence & Digital E, Shenzhen, Peoples R China
[2] Shenzhen Univ, Coll Comp Sci & Software Engn, Shenzhen, Peoples R China
[3] Hong Kong Baptist Univ, Dept Comp Sci, Hong Kong, Peoples R China
[4] Xiamen Univ, Sch Informat, Fujian Key Lab Sensing & Comp Smart City, Xiamen, Peoples R China
关键词
D O I
暂无
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
The imbalanced distribution of long-tailed data presents a considerable challenge for deep learning models, as it causes them to prioritize the accurate classification of head classes but largely disregard tail classes. The biased decision boundary caused by inadequate semantic information in tail classes is one of the key factors contributing to their low recognition accuracy. To rectify this issue, we propose to augment tail classes by grafting the diverse semantic information from head classes, referred to as head-to-tail fusion (H2T). We replace a portion of feature maps from tail classes with those belonging to head classes. These fused features substantially enhance the diversity of tail classes. Both theoretical analysis and practical experimentation demonstrate that H2T can contribute to a more optimized solution for the decision boundary. We seamlessly integrate H2T in the classifier adjustment stage, making it a plug-and-play module. Its simplicity and ease of implementation allow for smooth integration with existing long-tailed recognition methods, facilitating a further performance boost. Extensive experiments on various long-tailed benchmarks demonstrate the effectiveness of the proposed H2T. The source code is available at https://github.com/Keke921/H2T.
引用
收藏
页码:13581 / 13589
页数:9
相关论文
共 50 条
  • [31] Long-Tailed Partial Label Learning by Head Classifier and Tail Classifier Cooperation
    Jia, Yuheng
    Peng, Xiaorui
    Wang, Ran
    Zhang, Min-Ling
    THIRTY-EIGHTH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE, VOL 38 NO 11, 2024, : 12857 - 12865
  • [32] Class-Difficulty Based Methods for Long-Tailed Visual Recognition
    Saptarshi Sinha
    Hiroki Ohashi
    Katsuyuki Nakamura
    International Journal of Computer Vision, 2022, 130 : 2517 - 2531
  • [33] Prototype-based classifier learning for long-tailed visual recognition
    Xiu-Shen Wei
    Shu-Lin Xu
    Hao Chen
    Liang Xiao
    Yuxin Peng
    Science China Information Sciences, 2022, 65
  • [34] Long-tailed Visual Recognition via Gaussian Clouded Logit Adjustment
    Li, Mengke
    Cheung, Yiu-Ming
    Lu, Yang
    2022 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR 2022), 2022, : 6919 - 6928
  • [35] Rebalanced supervised contrastive learning with prototypes for long-tailed visual recognition
    Chang, Xuhui
    Zhai, Junhai
    Qiu, Shaoxin
    Sun, Zhengrong
    COMPUTER VISION AND IMAGE UNDERSTANDING, 2025, 252
  • [36] Long-tailed visual recognition with deep models: A methodological survey and evaluation
    Fu, Yu
    Xiang, Liuyu
    Zahid, Yumna
    Ding, Guiguang
    Mei, Tao
    Shen, Qiang
    Han, Jungong
    NEUROCOMPUTING, 2022, 509 : 290 - 309
  • [37] Contrastive dual-branch network for long-tailed visual recognition
    Miao, Jie
    Zhai, Junhai
    Han, Ling
    PATTERN ANALYSIS AND APPLICATIONS, 2025, 28 (01)
  • [38] Prototype-based classifier learning for long-tailed visual recognition
    Xiu-Shen WEI
    Shu-Lin XU
    Hao CHEN
    Liang XIAO
    Yuxin PENG
    Science China(Information Sciences), 2022, 65 (06) : 62 - 76
  • [39] Class-Difficulty Based Methods for Long-Tailed Visual Recognition
    Sinha, Saptarshi
    Ohashi, Hiroki
    Nakamura, Katsuyuki
    INTERNATIONAL JOURNAL OF COMPUTER VISION, 2022, 130 (10) : 2517 - 2531
  • [40] Prototype-based classifier learning for long-tailed visual recognition
    Wei, Xiu-Shen
    Xu, Shu-Lin
    Chen, Hao
    Xiao, Liang
    Peng, Yuxin
    SCIENCE CHINA-INFORMATION SCIENCES, 2022, 65 (06)