计算机工程 ›› 2020, Vol. 46 ›› Issue (2): 250-254,261.doi: 10.19678/j.issn.1000-3428.0053842

• 图形图像处理 • 上一篇    下一篇

基于尺度自适应卷积神经网络的人群计数算法

翟强, 王陆洋, 殷保群, 彭思凡, 邢思思   

  1. 中国科学技术大学 自动化系, 合肥 230027
  • 收稿日期:2019-01-28 修回日期:2019-03-05 发布日期:2019-03-14
  • 作者简介:翟强(1992-),男,硕士研究生,主研方向为图像处理;王陆洋,博士研究生;殷保群,教授;彭思凡、邢思思,硕士研究生。
  • 基金项目:
    装备预研领域基金(61403120201)。

Crowd Counting Algorithm Based on Scale Adaptive Convolutional Neural Network

ZHAI Qiang, WANG Luyang, YIN Baoqun, PENG Sifan, XING Sisi   

  1. Department of Automation, University of Science and Technology of China, Hefei 230027, China
  • Received:2019-01-28 Revised:2019-03-05 Published:2019-03-14

摘要: 为解决单幅图像中的人群遮挡和尺度变化问题,提出一种基于多列卷积神经网络的人群计数算法。利用具有不同尺寸感受野的卷积神经网络(CNN)和特征注意力模块自适应提取多尺度人群特征,引入可变形卷积增强CNN网络空间几何形变学习能力并优化特征图,从而生成高质量的密度图。Shanghai Tech和UCF_CC_50数据集上的实验结果表明,该算法能学习输入图和人群密度图之间的映射关系,且计数准确性高、鲁棒性强。

关键词: 人群计数, 卷积神经网络, 可变形卷积, 特征图, 密度图

Abstract: In order to solve the problem of crowd occlusion and scale change in a single image,this paper proposes a crowd counting algorithm based on multi-column convolution neural network.The algorithm uses Convolutional Neural Network(CNN) with receptive fields of different sizes and the feature attention module to adaptively extract multi-scale crowd features.The deformable convolution is introduced to enhance the learning ability of spatial geometric deformation of the network and optimize the feature map,so as to generate a high quality density map.Experimental results on the Shanghai Tech and UCF_CC_50 datasets show that the algorithm can learn the mapping relationship between input images and crowd density maps,and has high counting accuracy and robustness.

Key words: crowd counting, Convolutional Neural Network(CNN), deformable convolution, feature map, density map

中图分类号: