Research Article
An Improved Crowd Counting Method Based on YOLOv3
@INPROCEEDINGS{10.1007/978-3-031-04409-0_31, author={Shuang Zheng and Junfeng Wu and Fugang Liu and Yunhao Liang and Lingfei Zhao}, title={An Improved Crowd Counting Method Based on YOLOv3}, proceedings={Machine Learning and Intelligent Communications. 6th EAI International Conference, MLICOM 2021, Virtual Event, November 2021, Proceedings}, proceedings_a={MLICOM}, year={2022}, month={5}, keywords={Crowd density Target detection Convolutional neural network YOLOv3}, doi={10.1007/978-3-031-04409-0_31} }- Shuang Zheng
Junfeng Wu
Fugang Liu
Yunhao Liang
Lingfei Zhao
Year: 2022
An Improved Crowd Counting Method Based on YOLOv3
MLICOM
Springer
DOI: 10.1007/978-3-031-04409-0_31
Abstract
This paper proposes a method of crowd counting. We use ResNeSt-50 as the backbone network of YOLOv3. After the backbone network, we add SPP (Spatial Pyramid Potential) and PANet (Path Aggregation Network) to enhance the receptive field of convolutional neural network and improve the accuracy of stream of people or crowd counting in real application scenarios. In the application scenario of high-density crowd counting, an improved VGG network is used to design a deep network to capture high-level semantic information. At the same time, a shallow network is constructed to detect the head blob of people far away from the camera. The deep network and the shallow network are combined to detect high-density crowd. Finally, through the effective fusion of the above two network models, the accuracy and applicability of the algorithm are further improved. It can improve the detection accuracy in the case of small number of people and occlusion, and effectively reduce the estimation error in the scene with high density crowd.
