Embodied Crowd Counting
Occlusion is one of the fundamental challenges in crowd counting. In the community, various data-driven approaches have been developed to address this issue, yet their effectiveness is limited. This is mainly because most existing crowd counting datasets on which the methods are trained are based on...
Tallennettuna:
| Päätekijät: | , , , , , , , |
|---|---|
| Aineistotyyppi: | Artikkeli |
| Julkaistu: |
2025
|
| Aiheet: | |
| Linkit: | https://arxiv.org/abs/2503.08367 http://arxiv.org/abs/2503.08367 |
| Tagit: |
Ei tageja, Lisää ensimmäinen tagi!
|
| _version_ | 1879177978096648192 |
|---|---|
| author | Long, Runling Wang, Yunlong Wan, Jia Deng, Xiang Zhu, Xinting Guan, Weili Chan, Antoni B. Nie, Liqiang |
| author_browse | Chan, Antoni B. Deng, Xiang Guan, Weili Long, Runling Nie, Liqiang Wan, Jia Wang, Yunlong Zhu, Xinting |
| author_facet | Long, Runling Wang, Yunlong Wan, Jia Deng, Xiang Zhu, Xinting Guan, Weili Chan, Antoni B. Nie, Liqiang |
| author_sort | Long, Runling |
| building | arXiv |
| collection | Acceso abierto |
| description | Occlusion is one of the fundamental challenges in crowd counting. In the community, various data-driven approaches have been developed to address this issue, yet their effectiveness is limited. This is mainly because most existing crowd counting datasets on which the methods are trained are based on passive cameras, restricting their ability to fully sense the environment. Recently, embodied navigation methods have shown significant potential in precise object detection in interactive scenes. These methods incorporate active camera settings, holding promise in addressing the fundamental issues in crowd counting. However, most existing methods are designed for indoor navigation, showing unknown performance in analyzing complex object distribution in large scale scenes, such as crowds. Besides, most existing embodied navigation datasets are indoor scenes with limited scale and object quantity, preventing them from being introduced into dense crowd analysis. Based on this, a novel task, Embodied Crowd Counting (ECC), is proposed. We first build up an interactive simulator, Embodied Crowd Counting Dataset (ECCD), which enables large scale scenes and large object quantity. A prior probability distribution that approximates realistic crowd distribution is introduced to generate crowds. Then, a zero-shot navigation method (ZECC) is proposed. This method contains a MLLM driven coarse-to-fine navigation mechanism, enabling active Z-axis exploration, and a normal-line-based crowd distribution analysis method for fine counting. Experimental results against baselines show that the proposed method achieves the best trade-off between counting accuracy and navigation cost. |
| first_indexed | 2026-10-16T01:09:25Z |
| format | Article |
| fullrecord | <oai_dc:dc xmlns:oai_dc="http://www.openarchives.org/OAI/2.0/oai_dc/" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://www.openarchives.org/OAI/2.0/oai_dc/ https://www.openarchives.org/OAI/2.0/oai_dc.xsd"><identifier>arxiv-2503.08367</identifier><datestamp>2025-11-26</datestamp>
<dc:title>Embodied Crowd Counting</dc:title>
<dc:creator>Long, Runling</dc:creator>
<dc:creator>Wang, Yunlong</dc:creator>
<dc:creator>Wan, Jia</dc:creator>
<dc:creator>Deng, Xiang</dc:creator>
<dc:creator>Zhu, Xinting</dc:creator>
<dc:creator>Guan, Weili</dc:creator>
<dc:creator>Chan, Antoni B.</dc:creator>
<dc:creator>Nie, Liqiang</dc:creator>
<dc:subject>Computer Vision and Pattern Recognition</dc:subject>
<dc:description>Occlusion is one of the fundamental challenges in crowd counting. In the community, various data-driven approaches have been developed to address this issue, yet their effectiveness is limited. This is mainly because most existing crowd counting datasets on which the methods are trained are based on passive cameras, restricting their ability to fully sense the environment. Recently, embodied navigation methods have shown significant potential in precise object detection in interactive scenes. These methods incorporate active camera settings, holding promise in addressing the fundamental issues in crowd counting. However, most existing methods are designed for indoor navigation, showing unknown performance in analyzing complex object distribution in large scale scenes, such as crowds. Besides, most existing embodied navigation datasets are indoor scenes with limited scale and object quantity, preventing them from being introduced into dense crowd analysis. Based on this, a novel task, Embodied Crowd Counting (ECC), is proposed. We first build up an interactive simulator, Embodied Crowd Counting Dataset (ECCD), which enables large scale scenes and large object quantity. A prior probability distribution that approximates realistic crowd distribution is introduced to generate crowds. Then, a zero-shot navigation method (ZECC) is proposed. This method contains a MLLM driven coarse-to-fine navigation mechanism, enabling active Z-axis exploration, and a normal-line-based crowd distribution analysis method for fine counting. Experimental results against baselines show that the proposed method achieves the best trade-off between counting accuracy and navigation cost. </dc:description>
<dc:date>2025-03-11</dc:date>
<dc:date>2025-11-25</dc:date>
<dc:type>text</dc:type>
<dc:identifier>https://arxiv.org/abs/2503.08367</dc:identifier>
<dc:identifier>http://arxiv.org/abs/2503.08367</dc:identifier>
</oai_dc:dc> |
| id | arxiv-2503.08367 |
| institution | arXiv |
| last_indexed | 2026-10-16T01:09:25Z |
| publishDate | 2025 |
| publishDateRange | 2025 |
| publishDateSort | 2025 |
| record_format | oai_dc |
| spelling | arxiv-2503.083672025-11-26 Embodied Crowd Counting Long, Runling Wang, Yunlong Wan, Jia Deng, Xiang Zhu, Xinting Guan, Weili Chan, Antoni B. Nie, Liqiang Computer Vision and Pattern Recognition Occlusion is one of the fundamental challenges in crowd counting. In the community, various data-driven approaches have been developed to address this issue, yet their effectiveness is limited. This is mainly because most existing crowd counting datasets on which the methods are trained are based on passive cameras, restricting their ability to fully sense the environment. Recently, embodied navigation methods have shown significant potential in precise object detection in interactive scenes. These methods incorporate active camera settings, holding promise in addressing the fundamental issues in crowd counting. However, most existing methods are designed for indoor navigation, showing unknown performance in analyzing complex object distribution in large scale scenes, such as crowds. Besides, most existing embodied navigation datasets are indoor scenes with limited scale and object quantity, preventing them from being introduced into dense crowd analysis. Based on this, a novel task, Embodied Crowd Counting (ECC), is proposed. We first build up an interactive simulator, Embodied Crowd Counting Dataset (ECCD), which enables large scale scenes and large object quantity. A prior probability distribution that approximates realistic crowd distribution is introduced to generate crowds. Then, a zero-shot navigation method (ZECC) is proposed. This method contains a MLLM driven coarse-to-fine navigation mechanism, enabling active Z-axis exploration, and a normal-line-based crowd distribution analysis method for fine counting. Experimental results against baselines show that the proposed method achieves the best trade-off between counting accuracy and navigation cost. 2025-03-11 2025-11-25 text https://arxiv.org/abs/2503.08367 http://arxiv.org/abs/2503.08367 |
| spellingShingle | Computer Vision and Pattern Recognition Long, Runling Wang, Yunlong Wan, Jia Deng, Xiang Zhu, Xinting Guan, Weili Chan, Antoni B. Nie, Liqiang Embodied Crowd Counting |
| title | Embodied Crowd Counting |
| title_full | Embodied Crowd Counting |
| title_fullStr | Embodied Crowd Counting |
| title_full_unstemmed | Embodied Crowd Counting |
| title_short | Embodied Crowd Counting |
| title_sort | embodied crowd counting |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2503.08367 http://arxiv.org/abs/2503.08367 |
| work_keys_str_mv | AT longrunling embodiedcrowdcounting AT wangyunlong embodiedcrowdcounting AT wanjia embodiedcrowdcounting AT dengxiang embodiedcrowdcounting AT zhuxinting embodiedcrowdcounting AT guanweili embodiedcrowdcounting AT chanantonib embodiedcrowdcounting AT nieliqiang embodiedcrowdcounting |