Premium
One‐shot video‐based person re‐identification with variance subsampling algorithm
Author(s) -
Zhao Jing,
Yang Wenjing,
Yang Mingliang,
Huang Wanrong,
Yang Qiong,
Zhang Hongguang
Publication year - 2020
Publication title -
computer animation and virtual worlds
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.225
H-Index - 49
eISSN - 1546-427X
pISSN - 1546-4261
DOI - 10.1002/cav.1964
Subject(s) - computer science , variance (accounting) , credibility , sampling (signal processing) , task (project management) , identification (biology) , artificial intelligence , feature (linguistics) , quality (philosophy) , algorithm , sample (material) , pattern recognition (psychology) , machine learning , data mining , computer vision , linguistics , philosophy , botany , chemistry , accounting , management , filter (signal processing) , epistemology , chromatography , political science , economics , law , business , biology
Previous works propose the distance‐based sampling for unlabeled datapoints to address the few‐shot person re‐identification task, however, many selected samples may be assigned with wrong labels due to poor feature quality in these works, which negatively affects the learning procedure. In this article, we propose a novel sampling strategy to improve the quality of assigned pseudo‐labels, thus promoting the final performance. To illustrate, we first propose the concept of variance confidence to measure the credibility of pseudo‐labels, then we apply a novel variance subsampling algorithm to improve the accuracy of the selected sample labels. Our method combines distance confidence and variance confidence as a two‐round sampling criterion. Meanwhile, a variation decay strategy is used in our sampling process in combination with the actual distribution of features. We evaluate our approach on two publicly available datasets, MARS and DukeMTMC‐VideoReID, and achieve state‐of‐the‐art one‐shot performance.