CALM-Net: Curvature-Aware LiDAR Point Cloud-based Multi-Branch Neural Network for Vehicle Re-Identification

2025-10-16

AI Summary

To address the weak discriminability of geometric features and insufficient contextual information exploitation in 3D point cloud vehicle re-identification, this paper proposes a multi-branch neural network architecture integrating EdgeConv, point-wise attention, and curvature embedding. EdgeConv captures local structural relationships; the curvature embedding module explicitly models surface variation to enhance geometric sensitivity; and point-wise attention strengthens responses at salient points. These three branches extract complementary structural, geometric, and contextual features, respectively, which are adaptively fused to improve representation robustness. Evaluated on the nuScenes dataset, our method achieves a 1.97 percentage-point improvement in mAP over the strongest baseline. This demonstrates the effectiveness and advancement of curvature-guided, multi-granularity feature learning for 3D point cloud vehicle re-identification.

Technology Category

• Computer Vision: 3D Computer Vision
• Machine Learning: Feature Construction/Reformulation
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning

Application Category

• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

This paper presents CALM-Net, a curvature-aware LiDAR point cloud-based multi-branch neural network for vehicle re-identification. The proposed model addresses the challenge of learning discriminative and complementary features from three-dimensional point clouds to distinguish between vehicles. CALM-Net employs a multi-branch architecture that integrates edge convolution, point attention, and a curvature embedding that characterizes local surface variation in point clouds. By combining these mechanisms, the model learns richer geometric and contextual features that are well suited for the re-identification task. Experimental evaluation on the large-scale nuScenes dataset demonstrates that CALM-Net achieves a mean re-identification accuracy improvement of approximately 1.97% points compared with the strongest baseline in our study. The results confirms the effectiveness of incorporating curvature information into deep learning architectures and highlight the benefit of multi-branch feature learning for LiDAR point cloud-based vehicle re-identification.

Problem

• Learning discriminative features from 3D point clouds for vehicle re-identification
• Integrating curvature embedding to characterize local surface variations
• Improving re-identification accuracy using multi-branch neural network architecture

Innovation

• Multi-branch neural network for vehicle re-identification
• Integrates edge convolution with point attention
• Embeds curvature to characterize surface variation