Multi-view Structural Convolution Network for Domain-Invariant Point Cloud Recognition of Autonomous Vehicles

2025-01-27

AI Summary

To address domain shifts across sensors and scenes in point cloud perception for autonomous driving, this paper proposes the Multi-View Structured Convolutional Network (MSCN) for parameter-free learning of domain-invariant features. Methodologically, MSCN introduces two novel components: (i) a Structured Convolutional Layer (SCL) and a Structured Aggregation Layer (SAL) to explicitly model local geometric structures in point clouds; and (ii) a multi-view geometric feature extraction module coupled with a source-driven unseen-domain point cloud generation mechanism to enhance domain generalization. Evaluated on heterogeneous benchmarks—including Waymo Open Dataset, nuScenes, and SemanticKITTI—MSCN significantly mitigates domain shift, achieving an average 8.3% improvement in cross-domain recognition accuracy while maintaining robustness across diverse LiDAR configurations. To the best of our knowledge, this is the first work to achieve end-to-end, strongly generalizable, domain-invariant representation learning for point cloud recognition.

Technology Category

• Machine Learning: Transfer, Domain Adaptation, Multi-Task Learning
• Computer Vision: Multi-modal Vision
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• Semantics and Knowledge: Scalable techniques for the creation, curation, publication, maintenance, and consumption of large, Web-based, structured, reusable, knowledge graphs and ontologies

Abstract

Point cloud representation has recently become a research hotspot in the field of computer vision and has been utilized for autonomous vehicles. However, adapting deep learning networks for point cloud data recognition is challenging due to the variability in datasets and sensor technologies. This variability underscores the necessity for adaptive techniques to maintain accuracy under different conditions. In this paper, we present the Multi-View Structural Convolution Network (MSCN) designed for domain-invariant point cloud recognition. MSCN comprises Structural Convolution Layers (SCL) that extract local context geometric features from point clouds and Structural Aggregation Layers (SAL) that extract and aggregate both local and overall context features from point clouds. Additionally, our MSCN enhances feature representation robustness by training with unseen domain point clouds derived from source domain point clouds. This method acquires domain-invariant features and exhibits robust, consistent performance across various point cloud datasets, ensuring compatibility with diverse sensor configurations without the need for parameter adjustments. This highlights MSCN's potential to significantly improve the reliability and domain invariant features in different environments. Our code is available at https://github.com/MLMLab/MSCN.

Problem

• Adaptive Technology
• Point Cloud Recognition
• Autonomous Vehicles

Innovation

• Multi-View Structured Convolutional Network (MSCN)
• Structure Convolution Layer (SCL) and Structure Aggregation Layer (SAL)
• Unseen Point Cloud Data Training