Mixture of counting CNNs
Mixture of counting CNNs
复制标题
DOI：
10.1007/s00138-018-0955-6
复制
发表时间：
2017-03
影响因子：
3.3
通讯作者：
Shohei Kumagai;K. Hotta;Takio Kurita
中科院分区：
文献类型：
--
作者：
Shohei Kumagai;K. Hotta;Takio Kurita
本文提出了一种人群计数方法。由于密度和尺度的变化会导致目标的外观发生显著变化，所以人群计数比较困难。传统的人群计数方法通常使用一个预测器(例如，回归和多类分类器)。然而，这种只有一个预测器不能很好地统计具有显著外观变化的目标。本文提出了一种利用特定外观的多个卷积神经网络(CNN)来预测目标数目的方法，并根据测试图像的外观自适应地选择这些CNN。通过对选取的CNN进行整合，该方法对较大的外观变化具有较强的鲁棒性。在实验中，我们证实了该方法能够以比VGGNet、固定权值CNN集成和常规计数方法更低的计数误差来计数人群。此外，我们还证实，通过对CNN的训练，每个CNN都会自动对特定的人群外观(如密集区域和稀疏区域)进行专门化。
This paper proposes a crowd counting method. Crowd counting is difficult because of significant appearance changes of a target which caused by density and scale changes. Conventional crowd counting methods commonly utilize one predictor (e.g., regression and multi-class classifier). However, such only one predictor can not count targets with significant appearance changes well. In this paper, we propose to predict the number of targets using multiple convolutional neural networks (CNNs) specialized to a specific appearance, and those CNNs are adaptively selected according to the appearance of a test image. By integrating the selected CNNs, the proposed method has the robustness to large appearance changes. In experiments, we confirm that the proposed method can count crowd with lower counting error than VGGNet, integration of CNNs with fixed weights and conventional counting methods. Moreover, we confirm that each CNN automatically specialized to a specific appearance (e.g., dense region and sparse region) of crowd through training of CNNs.