Prototype-based classifier learning for long-tailed visual recognition  

作　　者：Xiu-Shen WEI Shu-Lin XU Hao CHEN Liang XIAO Yuxin PENG 

机构地区：[1]School of Computer Science and Engineering,Nanjing University of Science and Technology,Nanjing 210094,China [2]State Key Laboratory of Integrated Services Networks,Xidian University,Xi'an 710071,China [3]Key Lab of Intelligent Perception and Systems for High-Dimensional Information of Ministry of Education,and Jiangsu Key Lab of Image and Video Understanding for Social Security,Nanjing 210094,China [4]Wangxuan Institute of Computer Technology,Peking University,Beijing 100871,China

出　　处：《Science China(Information Sciences)》2022年第6期58-72,共15页中国科学（信息科学）（英文版）

基　　金：supported by National Key R&D Program of China(Grant No.2021YFA1001100);National Natural Science Foundation of China(Grant Nos.61925201,62132001,U21B2025,61871226);Natural Science Foundation of Jiangsu Province of China(Grant No.BK20210340);Fundamental Research Funds for the Central Universities(Grant No.30920041111);CAAI-Huawei MindSpore Open Fund,and Beijing Academy of Artificial Intelligence(BAAI)。

摘　　要：In this paper,we tackle the long-tailed visual recognition problem from the categorical prototype perspective by proposing a prototype-based classifier learning(PCL)method.Specifically,thanks to the generalization ability and robustness,categorical prototypes reveal their advantages of representing the category semantics.Coupled with their class-balance characteristic,categorical prototypes also show potential for handling data imbalance.In our PCL,we propose to generate the categorical classifiers based on the prototypes by performing a learnable mapping function.To further alleviate the impact of imbalance on classifier generation,two kinds of classifier calibration approaches are designed from both prototype-level and example-level aspects.Extensive experiments on five benchmark datasets,including the large-scale iNaturalist,Places-LT,and ImageNet-LT,justify that the proposed PCL can outperform state-of-the-arts.Furthermore,validation experiments can demonstrate the effectiveness of tailored designs in PCL for long-tailed problems.

关 键 词：long-tailed distribution categorical prototype classifier generation classifier calibration class imbalance 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] TP18[自动化与计算机技术—计算机科学与技术]