DSGC-Net: A Dual-Stream Graph Convolutional Network for Crowd Counting via Feature Correlation Mining

2025-09-02

AI Summary

In complex crowd scenes, uneven density distributions and variations in viewpoint and pose lead to inconsistent individual representations. To address this, we propose a Dual-Stream Graph Convolutional Network (Dual-Stream GCN) that innovatively constructs two complementary semantic graph structures: a density-driven graph modeling spatial density correlations, and a representation-driven graph leveraging global feature similarity to learn pose-robust individual representations. Our method jointly integrates a density prediction module with graph convolutional semantic relation modeling, significantly enhancing adaptability to multi-scale density variations and geometric deformations. Evaluated on ShanghaiTech Part A and Part B, our approach achieves mean absolute errors (MAE) of 48.9 and 5.9, respectively—outperforming state-of-the-art methods. These results validate the effectiveness and generalizability of dual-semantic graph modeling for crowd counting.

Technology Category

• Machine Learning: Graph-based Machine Learning
• Computer Vision: Visual Reasoning & Symbolic Representations
• Humans and AI: Crowd Sourcing and Human Computation

Application Category

• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Security and Privacy: Large-scale security measurements

Abstract

Deep learning-based crowd counting methods have achieved remarkable progress in recent years. However, in complex crowd scenarios, existing models still face challenges when adapting to significant density distribution differences between regions. Additionally, the inconsistency of individual representations caused by viewpoint changes and body posture differences further limits the counting accuracy of the models. To address these challenges, we propose DSGC-Net, a Dual-Stream Graph Convolutional Network based on feature correlation mining. DSGC-Net introduces a Density Approximation (DA) branch and a Representation Approximation (RA) branch. By modeling two semantic graphs, it captures the potential feature correlations in density variations and representation distributions. The DA branch incorporates a density prediction module that generates the density distribution map, and constructs a density-driven semantic graph based on density similarity. The RA branch establishes a representation-driven semantic graph by computing global representation similarity. Then, graph convolutional networks are applied to the two semantic graphs separately to model the latent semantic relationships, which enhance the model's ability to adapt to density variations and improve counting accuracy in multi-view and multi-pose scenarios. Extensive experiments on three widely used datasets demonstrate that DSGC-Net outperforms current state-of-the-art methods. In particular, we achieve MAE of 48.9 and 5.9 in ShanghaiTech Part A and Part B datasets, respectively. The released code is available at: https://github.com/Wu-eon/CrowdCounting-DSGCNet.

Problem

• Addressing density distribution differences in crowd counting
• Improving individual representation consistency across viewpoints
• Enhancing counting accuracy in multi-view and multi-pose scenarios

Innovation

• Dual-stream network with density and representation branches
• Graph convolutional networks model feature correlations
• Semantic graphs capture density and representation similarities