Self-calibrated region-level regression for crowd counting  

作　　者：Jiawen ZHU Wenda ZHAO You HE Huchuan LU 

机构地区：[1]School of Information and Communication Engineering,Dalian University of Technology,Dalian 116024,China [2]Department of Electronic Engineering,Tsinghua University,Beijing 100084,China

出　　处：《Science China(Information Sciences)》2025年第4期383-384,共2页中国科学(信息科学)(英文版)

基　　金：supported by National Natural Science Foundation of China(Grant Nos.62176038,U1903215)。

摘　　要：Accurate crowd counting in natural images has become increasingly attractive owing to its numerous real-world applications,e.g.,crowd analysis and video surveillance.Despite significant progress in crowd counting[1,2],challenges(such as scale variation and background clutter)remain.To fully utilize spatial information,existing crowd counting approaches[3,4]mainly estimate a density map,where point annotations are smoothed via a Gaussian kernel to generate probabilities indicating the presence of a crowd.

关 键 词：crowd counting natural images point annotations smoothed estimate density mapwhere video surveillancedespite self calibrated gaussian kernel crowd counting challenges such 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]