Dense crowd counting based on adaptive scene division
- PDF / 3,612,035 Bytes
- 12 Pages / 595.276 x 790.866 pts Page_size
- 9 Downloads / 208 Views
ORIGINAL ARTICLE
Dense crowd counting based on adaptive scene division Ying Yu1 · Huilin Zhu1 · Lewei Wang1 · Witold Pedrycz2,3 Received: 27 March 2020 / Accepted: 20 September 2020 © Springer-Verlag GmbH Germany, part of Springer Nature 2020
Abstract With the rapid development of computer vision and artificial intelligence, crowd counting has attracted significant attention from researchers and many well-known methods were proposed. However, due to interocclusions, perspective distortion, and uneven crowd distribution, crowd counting is still a highly challenging task in crowd analysis. Motivated by granular computing, a novel end-to-end crowd counting network (GrCNet) is proposed to enable the problem of crowd counting to be conceptualized at different levels of granularity, and to map problem into computationally tractable subproblems. It shows that by adaptively dividing the image into granules and then feeding the granules into different counting subnetworks separately, the scale variation range of image is narrowed and the the adaptability of counting algorithm to different scenarios is improved. Experiments on four well-known crowd counting benchmark datasets indicate that GrCNet achieves state-ofthe-art counting performance and high robustness in dense crowd counting. Keywords Crowd counting · Granular computing · Density map · Feature extraction · Dilated convolution
1 Introduction Crowd counting is a fundamental task of crowd analysis. It aims to estimate the number of individuals in a sparse or dense crowd scene. With the rapid urbanization around the worldwide, the urban population is growing rapidly. Exponential growth in the urban population has led to an increased number of activities such as vocal concert, sporting events, political rallies, etc., thereby resulting in more frequent crowd gatherings in the recent years. In such scenarios, it is essential to count the number of individuals in a crowded scene for better management, safety and security [1]. Consequently, crowd counting has emerged as a crucial focus in crowd analysis for providing valuable information to anticipate overcrowding or detect the abnormal events. This endeavour is also further motivated by the need for a sophisticated crowd analysis system. * Ying Yu [email protected] 1
College of Software, East China Jiaotong University, Nanchang 330013, China
2
Department of Electrical and Computer Engineering, University of Alberta, Edmonton, AB T6G 2G7, Canada
3
System Research Institute, Polish Academy of Sciences, Warsaw, Poland
Crowd counting has a variety of real-world applications, such as public safety management [2, 3], intelligent surveillance [4], and urban planning [5]. The methods developed for crowd counting can be easily extended to object counting tasks in many other domains, such as vehicle counting [6, 7], animal counting [8], etc. With the rapid development of computer vision and artificial intelligence technology, crowd counting has attracted significant attention from researches in the recent past a
Data Loading...
