Semantic-aware Transmission for Robust Point Cloud Classification

Tianxiao Han,Kaiyi Chi,Qianqian Yang,Zhiguo Shi

2023-06-23

Abstract:As three-dimensional (3D) data acquisition devices become increasingly prevalent, the demand for 3D point cloud transmission is growing. In this study, we introduce a semantic-aware communication system for robust point cloud classification that capitalizes on the advantages of pre-trained Point-BERT models. Our proposed method comprises four main components: the semantic encoder, channel encoder, channel decoder, and semantic decoder. By employing a two-stage training strategy, our system facilitates efficient and adaptable learning tailored to the specific classification tasks. The results show that the proposed system achieves classification accuracy of over 89\% when SNR is higher than 10 dB and still maintains accuracy above 66.6\% even at SNR of 4 dB. Compared to the existing method, our approach performs at 0.8\% to 48\% better across different SNR values, demonstrating robustness to channel noise. Our system also achieves a balance between accuracy and speed, being computationally efficient while maintaining high classification performance under noisy channel conditions. This adaptable and resilient approach holds considerable promise for a wide array of 3D scene understanding applications, effectively addressing the challenges posed by channel noise.

Signal Processing

What problem does this paper attempt to address?

The paper aims to address two main issues in the transmission of three-dimensional (3D) point cloud data:

1. **Efficient Transmission**: With the widespread application of 3D data acquisition devices such as LiDAR and RGB-D cameras, the demand for 3D point cloud data is growing. However, the size of point cloud data is relatively large, which puts significant bandwidth pressure on standard communication systems. Therefore, the paper proposes a semantic-aware communication system to improve transmission efficiency.

2. **Robustness**: Although existing compression methods have improved transmission efficiency, they still fall short when facing channel noise. To overcome this challenge, the paper introduces a method based on the pre-trained Point-BERT model to effectively classify point cloud data and ensure high classification accuracy even under adverse channel conditions.

Specifically, the main contributions of the paper include:

- Proposing a system that contains four main components: a semantic encoder, a channel encoder, a channel decoder, and a semantic decoder. These components work together for the efficient processing and transmission of point cloud data.

- Utilizing Farthest Point Sampling (FPS) and K-Nearest Neighbors (KNN) algorithms for downsampling and sub-cloud generation to extract key information.

- Adopting the pre-trained Point-BERT model as the core part of the semantic encoder to extract semantic information relevant to specific classification tasks.

- Designing a two-stage training strategy that enables the system to adapt to different channel conditions and achieve efficient and adaptive learning.

Experimental results show that the proposed system exhibits significant superiority and robustness in point cloud classification tasks under different Signal-to-Noise Ratio (SNR) conditions, especially in low SNR environments. Moreover, compared to existing methods, this approach also achieves higher accuracy and lower compression ratios, proving its potential in practical applications.