Abstract | ||
---|---|---|
Crowd density estimation has important practical significance for effectively suppressing the occurrence of stampede accidents. However, the crowd counting task can be easily interfered by various factors such as perspective, congestion, occlusion, density, etc., which makes accurate crowd counting a challenging task. To solve these problems, in this paper, we propose an effective hierarchical aggregation module to fuse different scale information in the network. Since the crowd counting task is seriously interfered by the surrounding environment, in this paper we propose to use attention mechanism module to weight the spatial position of the network learned feature map to effectively limit the interference of the background region to the crowd counting task. Finally, a large number of related experiments show that our model in this paper has strong generalization ability while having better performance on several public datasets compared to existing model algorithms. |
Year | DOI | Venue |
---|---|---|
2019 | 10.1007/978-3-030-36808-1_39 | NEURAL INFORMATION PROCESSING (ICONIP 2019), PT IV |
Keywords | DocType | Volume |
Crowd counting, Effective hierarchical aggregation, Attention mechanism | Conference | 1142 |
ISSN | Citations | PageRank |
1865-0929 | 0 | 0.34 |
References | Authors | |
0 | 5 |
Name | Order | Citations | PageRank |
---|---|---|---|
Huimin Grace Guo | 1 | 13 | 2.68 |
Fujin He | 2 | 3 | 0.75 |
Xin Cheng | 3 | 0 | 0.34 |
Xinghao Ding | 4 | 591 | 52.95 |
Yue Huang | 5 | 317 | 29.82 |