Crowd counting via deep spectral adaptive modulation for dense scenes

LIU Meng; ZHANG Kaibing; MENG Yalei

doi:10.16152/j.cnki.xdxbzr.2026-03-010

您当前的位置：

首页 >

文章列表页 >

Crowd counting via deep spectral adaptive modulation for dense scenes

更新时间：2026-07-07

- Crowd counting via deep spectral adaptive modulation for dense scenes
- Journal of Northwest University (Natural Science Edition) Vol. 56, Issue 3, Pages: 572-582(2026)
- 作者机构：
  
  1.西安工程大学　计算机科学学院，陕西　西安　710048
  2.陕西省服装设计智能化重点实验室，陕西　西安　710048
- 作者简介：
- 基金信息：
- DOI：10.16152/j.cnki.xdxbzr.2026-03-010
  CLC： TP391
- Received：20 December 2025，
  
  Revised：2026-01-05，
  
  Published：25 June 2026
- 稿件说明：
移动端阅览
LIU Meng, ZHANG Kaibing, MENG Yalei. Crowd counting via deep spectral adaptive modulation for dense scenes[J]. Journal of Northwest University (Natural Science Edition), 2026, 56(3): 572-582.
DOI：

LIU Meng, ZHANG Kaibing, MENG Yalei. Crowd counting via deep spectral adaptive modulation for dense scenes[J]. Journal of Northwest University (Natural Science Edition), 2026, 56(3): 572-582. DOI： 10.16152/j.cnki.xdxbzr.2026-03-010.

摘要

在人群计数领域，由于场景中人群密度和遮挡程度高度变化，传统方法往往难以在保持局部细节和捕捉全局上下文之间取得平衡，导致在高拥挤度或透视畸变严重的图像上性能下降。针对这一问题，提出一种基于深度频谱自适应调制的密集场景人群计数方法（deep spectral adaptive modulation for dense scenes，DSAMNet），旨在同时提升密度图估计的空间分辨率与全局计数一致性。首先，通过深度感知模块联合提取多尺度纹理特征与粗略深度估计，并利用预测深度引导透视校正，实现几何一致的特征采样；然后，引入空间自适应高频编码模块，在编码过程中依据局部深度动态调制坐标频率，从而增强几何敏感区域的位置表达能力，提升网络对尺度与结构变化的适应性；最后，隐式密度解码模块通过跨域注意力机制融合视觉与几何表征，并借助分层条件化多层感知机进行逐点密度回归，实现对连续密度场的高质量重建。实验结果表明，所提方法在多个主流人群计数数据集上取得了优异性能，展现出在复杂透视场景下良好的鲁棒性与泛化能力。

Abstract

In crowd counting

traditional methods often struggle to balance preserving local details and capturing global context due to the high variability in crowd density and occlusion levels within scenes

leading to performance degradation in highly congested or severely perspective-distorted images. To address this issue

this paper proposes a dense-scene crowd counting method based on deep spectral adaptive modulation(DSAMNet)

aiming to simultaneously enhance the spatial resolution of density map estimation and global counting consistency. Specifically

the method first employs a depth-aware module to jointly extract multi-scale texture features and coarse depth estimates

utilizing predicted depth to guide perspective correction and achieve geometrically consistent feature sampling. Subsequently

a spatially adaptive high-frequency encoding module is introduced

which dynamically modulates coordinate frequencies based on local depth during the encoding process

thereby enhancing the positional representation capacity in geometrically sensitive regions and improving the network’s adaptability to scale and structural variations. Finally

an implicit density decoding module integrates visual and geometric representations through a cross-domain attention mechanism and performs point-wise density regression via a hierarchically conditioned multi-layer perceptron

achieving high-quality reconstruction of continuous density fields. Experimental results demonstrate that the proposed method achieves superior performance on several mainstream crowd counting datasets

exhibiting strong robustness and generalization capability in complex perspective scenarios.

关键词

Keywords

references

王怡婷 . 基于深度学习的人群计数方法研究 [D ] . 北京：北京建筑大学， 2024 ： 1 .

扶齐彦 . 基于信道感知和迁移学习的复杂交通场景中密集车辆计数 [D ] . 南昌：南昌大学， 2024 ： 2 .

陈慧 . 基于多维度感知特征金字塔的人群计数算法 [J ] . 南通职业大学学报， 2025 , 39 ( 1 ): 72 - 79 .

CHEN H . Crowd counting algorithm based on multi-dimensional perceptual feature pyramid [J ] . Journal of Nantong Vocational University , 2025 , 39 ( 1 ): 72 - 79 .

杨月，连淑娟，马朝鹏，等 . 基于HOD-Net模型的教室密集场景人群计数方法 [J/OL ] . 计算机应用与软件 .（ 2025-04-18 ）[ 2025-12-10 ] . https://kns.cnki.net/KCMS/detail/detail.aspx?filename=JYRJ20250417001&dbname=CJFD&dbcode=CJFQ https://kns.cnki.net/KCMS/detail/detail.aspx?filename=JYRJ20250417001&dbname=CJFD&dbcode=CJFQ .

YANG Y , LIAN S J , MA C P , et al . Dense crowd counting method for classroom scenes based on HOD-Net model [J/OL ] . Computer Applications and Software .( 2025-04-18 ) [ 2025-12-10 ] . https://kns.cnki.net/KCMS/detail/detail.aspx?filename=JYRJ20250417001&dbname=CJFD&dbcode=CJFQ https://kns.cnki.net/KCMS/detail/detail.aspx?filename=JYRJ20250417001&dbname=CJFD&dbcode=CJFQ .

时东亮，葛艳，徐慕君 . 基于多尺度特征加权融合注意力的密集人群计数算法 [J ] . 计算机系统应用， 2025 , 34 ( 3 ): 210 - 219 .

SHI D L , GE Y , XU M J . Dense crowd counting algorithm based on multi-scale feature weighted fusion attention [J ] . Computer Systems and Applications , 2025 , 34 ( 3 ): 210 - 219 .

文帅，蒋勇，杨丹，等 . 基于多尺度注意力网络的密集人群计数 [J ] . 计算机应用与软件， 2025 , 42 ( 1 ): 130 - 136 .

WEN S , JIANG Y , YANG D , et al . Dense crowd counting based on multi-scale attention network [J ] . Computer Applications and Software , 2025 , 42 ( 1 ): 130 - 136 .

郑棣文，石洋宇，谢承杰，等 . 多尺度感知与红外特征增强的RGB-T人群计数方法 [J/OL ] . 北京航空航天大学学报 .（ 2024-09-19 ）[ 2025-12-10 ] . https://link.cnki.net/urlid/11.2625.v.20240919.1254.003 https://link.cnki.net/urlid/11.2625.v.20240919.1254.003 .

ZHENG D W , SHI Y Y , XIE C J , et al . RGB-T crowd counting method with multi-scale perception and infrared feature enhancement [J/OL ] . Journal of Beihang University . ( 2024-09-19 ) [ 2025-12-10 ] . https://link.cnki.net/urlid/11.2625.v.20240919.1254.003 https://link.cnki.net/urlid/11.2625.v.20240919.1254.003 .

谢嘉豪 . 基于多尺度超分辨率信息增强的人群计数方法 [D ] . 北京：北京邮电大学， 2024 ： 18 - 19 .

MOLAEI A , AMINIMEHR A , TAVAKOLI A , et al . Implicit neural representation in medical imaging: A comparative survey [C ] // 2023 IEEE/CVF International Conference on Computer Vision Workshops (ICCVW) . Paris : IEEE , 2023 : 2373 - 2383 .

MILDENHALL B , SRINIVASAN P P , TANCIK M , et al . NeRF: Representing scenes as neural radiance fields for view synthesis [M ] // Computer Vision-ECCV 2020 . Cham : Springer Internatinal Publishing , 2020 : 405 - 421 .

PENG S Y , NIEMEYER M , MESCHEDER L , et al . Convolutional occupancy networks [C ] // Computer Vision-ECCV 2020 . Cham : Springer Internatinal Publishing , 2020 : 523 - 540 .

TANCIK M , SRINIVASAN P P , MILDENHALL B , et al . Fourier features let networks learn high frequency functions in low dimensional domains [EB/OL ] .( 2020-06-18 ) [ 2025-12-10 ] . https://arxiv.org/abs/2006.10739 https://arxiv.org/abs/2006.10739 .

LIN C H , MA W C , TORRALBA A , et al . BARF: Bundle-adjusting neural radiance fields [C ] // 2021 IEEE/CVF International Conference on Computer Vision . Montreal, QC : IEEE , 2021 : 5721 - 5731 .

RAN Y , ZENG J , HE S , et al . NeurAR: Neural uncertainty for autonomous 3D reconstruction with implicit neural representations [J ] . IEEE Robotics and Automation Letters , 2023 , 8 ( 2 ): 1125 - 1132 .

BARRON J T , MILDENHALL B , TANCIK M , et al . Mip-NeRF: A multiscale representation for anti-aliasing neural radiance fields [C ] // 2021 IEEE/CVF International Conference on Computer Vision . Montreal, QC : IEEE , 2021 : 5855 - 5864 .

ZHANG Y Y , ZHOU D S , CHEN S Q , et al . Single-image crowd counting via multi-column convolutional neural network [C ] // 2016 IEEE Conference on Computer Vision and Pattern Recognition . Las Vegas, NV : IEEE , 2016 : 589 - 597 .

IDREES H , TAYYAB M , ATHREY K , et al . Composition loss for counting, density map estimation and localization in dense crowds [M ] // Computer Vision-ECCV 2018 . Cham : Springer International Publishing , 2018 : 544 - 559 .

SINDAGI V A , YASARLA R , PATEL V M . JHUCROWD++: Large-scale crowd counting dataset and a benchmark method [J ] . IEEE Transactions on Pattern Analysis and Machine Intelligence , 2022 , 44 ( 5 ): 2594 - 2609 .

ZHAO L , DAI H , WU Z , et al . CP-CNN: Core-periphery principle guided convolutional neural network [EB/OL ] .( 2023-03-27 )[ 2025-12-10 ] . https://arxiv.org/abs/2304.10515 https://arxiv.org/abs/2304.10515 .

LI Y H , ZHANG X F , CHEN D M . CSRNet: Dilated convolutional neural networks for understanding the highly congested scenes [C ] // 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition . Salt Lake City, UT : IEEE , 2018 : 1091 - 1100 .

LIU W Z , SALZMANN M , FUA P . Context-aware crowd counting [C ] // 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition . Long Beach, CA : IEEE , 2019 : 5094 - 5103 .

WANG B , LIU H , SAMARAS D , et al . Distribution matching for crowd counting [EB/OL ] . ( 2020-09-28 ) [ 2025-12-10 ] . https://arxiv.org/abs/2009.13077 https://arxiv.org/abs/2009.13077 .

XIONG H P , LU H , LIU C X , et al . From open set to closed set: Supervised spatial divide-and-conquer for object counting [J ] . International Journal of Computer Vision , 2023 , 131 ( 7 ): 1722 - 1740 .

GUO X Y , GAO M L , ZHAI W Z , et al . Scale region recognition network for object counting in intelligent transportation system [J ] . IEEE Transactions on Intelligent Transportation Systems , 2023 , 24 ( 12 ): 15920 - 15929 .

CHEN J , GAO M , LI Q , et al . Privacy-aware crowd counting by decentralized learning with parallel transformers [J ] . Internet of Things , 2024 , 26 : 101167 .

WANG T , ZHANG T , ZHANG K B , et al . Context attention fusion network for crowd counting [J ] . Knowledge-Based Systems , 2023 , 271 : 110541 .

ZHAO Z Y , LI X L . Deformable density estimation via adaptive representation [J ] . IEEE Transactions on Image Processing , 2023 , 32 : 1134 - 1144 .

ABOUSAMRA S , HOAI M , SAMARAS D , et al . Localization in the crowd with topological constraints [J ] . Proceedings of the AAAI Conference on Artificial Intelligence , 2021 , 35 ( 2 ): 872 - 881 .

SAM D B , PERI S V , SUNDARARAMAN M N , et al . Locate, size, and count: Accurately resolving people in dense crowds via detection [J ] . IEEE Transactions on Pattern Analysis and Machine Intelligence , 2021 , 43 ( 8 ): 2739 - 2751 .

CAO X K , WANG Z P , ZHAO Y Y , et al . Scale aggregation network for accurate and efficient crowd counting [M ] // Computer Vision-ECCV 2018 . Cham : Springer Internatloral Publishing , 2018 : 757 - 773 .

WANG Q , GAO J Y , LIN W , et al . Learning from synthetic data for crowd counting in the wild [C ] // 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition . Long Beach, CA : IEEE , 2019 : 8190 - 8199 .

SONG Q Y , WANG C G , WANG Y B , et al . To choose or to fuse? Scale selection for crowd counting [J ] . Proceedings of the AAAI Conference on Artificial Intelligence , 2021 , 35 ( 3 ): 2576 - 2583 .

MA Z H , WEI X , HONG X P , et al . Bayesian loss for crowd count estimation with point supervision [C ] // 2019 IEEE/CVF International Conference on Computer Vision (ICCV) . Seoul, Korea (South) : IEEE , 2019 : 6141 - 6150 .

CHENG J A , LI Q L , CHEN J Y , et al . Efficient vehicular counting via privacy-aware aggregation network [J ] . Measurement Science and Technology , 2025 , 36 ( 2 ): 026213 .

Views

下载量

CSCD

Alert me when the article has been cited

提交

Tools

Publicity Resources

Attribute reduction in generalized multi-granularity formal decision contexts

Sedimentary facies characteristics and distribution patterns of the sub-member Ma 51 in the Majiagou Formation, eastern Yanchang exploration block, Ordos Basin

The influence of tourist-employee interaction on heritage protection behavior in intangible culture heritage exhibition area: A case study of Panxiu workshop in Huzhu Tu autonomous county

Road extraction from remote sensing images based on attention and adaptive feature fusion

The influence mechanism of the digital economy on the integration of culture and tourism in the Domestic Section of the Silk Road Economic Belt

Related Author

No data

Related Institution

No data

⁰