期刊
IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS
卷 16, 期 1, 页码 566-576出版社
IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
DOI: 10.1109/TII.2019.2935244
关键词
Convolutional neural network (CNN); cross-level and multiscale features; crowd counting; density map; scale aggregation network
类别
资金
- National Key Research and Development Program of China [2017YFC0820205]
- National Natural Science Foundation of China [61703198]
- Natural Science Foundation for Distinguished Young Scholars of Jiangxi Province [2018ACB21014]
- Open Fund of State Key Laboratory of Management and Control for Complex Systems [20180109]
Automated people counting in crowd scenes is challenging due to large variations in scale, density, and background clutter. To tackle them, we propose a novel cross-level parallel network (CLPNet) by extracting multiple low-level features from VGG16 and fusing them with specific scale aggregation modules in the high-level stage. To deal with scale variation, we design five different aggregation modules for multiscale fusion. Furthermore, the ground truth is processed skillfully to eliminate the mismatches caused by the scale variation between heads and density maps. To cope with background clutter, cross-level feature fusion is implemented. Higher-level semantic information could effectively separate head from background and regain the lost low-level detailed information. To address the variation of density, we design a parallel network, in which two separate channels focus on different density-level estimation, and attain more accurate counting results. Finally, we evaluate the proposed CLPNet on four representative crowd counting datasets, i.e., ShanghaiTech, UCF_CC_50, WorldExpo'10, and UCF_QNRF. The experimental results demonstrate that with the cross-level and multiscale structure CLPNet achieves superior performance compared with the state-of-the-art crowd counting methods.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据