Home // ICONS 2022, The Seventeenth International Conference on Systems // View article

CROWD SZ: A Large-scale Multi-view Crowd Counting Semantic Dataset

Authors:
Jiajia Wu
Yufeng Lin
Cheng Wu
Jin Zhang
Lijun Zhang

Keywords: Crowd counting; Semantic understanding; Data fusion

Abstract:
In recent years, there have been many large social gatherings and stampedes. High-density crowd counting and density estimation have become a research hotspot in the field of video surveillance. However, traditional datasets expose the limitations of a single perspective and limited crowd size, which cannot meet the research needs of a wide-area place. This paper proposes a new large-scale multi-view dataset, taking the urban life square near Jinji Lake in Suzhou city, Jiangsu Province, China as the research object. A single camera cannot cover the whole place, so we collect surveillance images from multiple perspectives. The low-altitude monitoring image has obvious human characteristics, while the high-altitude monitoring image provides the trend of crowd distribution. Combining these two kinds of information, the trend of crowd change can be predicted more accurately. This dataset is characterized by rapid crowd change in a short time, large aggregation scale and complex illumination conditions, which brings new challenges to crowd counting research.

Pages: 12 to 17

Copyright: Copyright (c) IARIA, 2022

Publication date: April 24, 2022

Published in: conference

ISSN: 2308-4243

ISBN: 978-1-61208-941-6

Location: Barcelona, Spain

Dates: from April 24, 2022 to April 28, 2022