Improving performance of deep learning models for 3D point cloud semantic segmentation via attention mechanisms

Vanian, Vazgen; Zamanakos, Georgios; Pratikakis, Ioannis

doi:10.1016/j.cag.2022.06.010

Cited by 19 publications

(11 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…As a result, the detection results are affected. We incorporate the large sparse kernel convolution and convolutional block attention module (CBAM) [36] to enhance the representational ability of the neural network. The CBAM focuses on our target regions of interest while suppressing information from irrelevant regions.…”

Section: It Obtains High-precision Detection Resultsmentioning

confidence: 99%

AFRNet: Anchor-Free Object Detection Using Roadside LiDAR in Urban Scenes

Wang,

Lan,

2024

Remote Sensing

View full text Add to dashboard Cite

In urban settings, roadside infrastructure LiDAR is a ground-based remote sensing system that collects 3D sparse point clouds for the traffic object detection of vehicles, pedestrians, and cyclists. Current anchor-free algorithms for 3D point cloud object detection based on roadside infrastructure face challenges related to inadequate feature extraction, disregard for spatial information in large 3D scenes, and inaccurate object detection. In this study, we propose AFRNet, a two-stage anchor-free detection network, to address the aforementioned challenges. We propose a 3D feature extraction backbone based on the large sparse kernel convolution (LSKC) feature set abstraction module, and incorporate the CBAM attention mechanism to enhance the large scene feature extraction capability and the representation of the point cloud features, enabling the network to prioritize the object of interest. After completing the first stage of center-based prediction, we propose a refinement method based on attentional feature fusion, where fused features incorporating raw point cloud features, voxel features, BEV features, and key point features are used for the second stage of refinement to complete the detection of 3D objects. To evaluate the performance of our detection algorithms, we conducted experiments using roadside LiDAR data from the urban traffic dataset DAIR-V2X, based on the Beijing High-Level Automated Driving Demonstration Area. The experimental results show that AFRNet has an average of 5.27 percent higher detection accuracy than CenterPoint for traffic objects. Comparative tests further confirm that our method achieves high accuracy in roadside LiDAR object detection.

show abstract

Section: It Obtains High-precision Detection Resultsmentioning

confidence: 99%

AFRNet: Anchor-Free Object Detection Using Roadside LiDAR in Urban Scenes

Wang,

Lan,

2024

Remote Sensing

View full text Add to dashboard Cite

show abstract

“…Hu et al 30 proposed ACU‐Net that includes 3D spatial attention blocks to enrich the spatial details and feature representation of lesions in the decoding stage. Vanian et al 31 improving DL models for 3D point cloud semantic segmentation via attention mechanism.…”

Section: Related Workmentioning

confidence: 99%

Efficient multiscale spatial attention 3D abdominal multiorgan segmentation model

Yan,

Hou,

Shen

et al. 2024

Int J Imaging Syst Tech

View full text Add to dashboard Cite

Accurate and efficient 3D medical image segmentation models are critical for clinical applications. However, many models which performed well in recent years have consumed significant computational cost, especially those based on self‐attention. For more efficient and accurate segmentation of medical 3D images, we develop an efficient multiscale spatial attention model for 3D abdominal multiorgan segmentation. We use 1D and 2D kernels in the encoder and decoder modules to simulate 3D kernels to reduce the number of parameters and computational cost. Furthermore, we use the lightweight spatial attention and mixed pooling module to capture cross‐dimensional information, multiscale features, and long‐range contextual information. We achieved a Dice Similarity Coefficient (DSC) of 89.86 and a Normalized Surface Dice (NSD) of 78.2 on the FLARE dataset with only 1183 M GPU memory usage, 9 M parameters, and 193G floating point operations(FLOPs).

show abstract

“…To validate the effectiveness of the CNN-SA network, a multi-source and multi-modal data fusion network based on the self-attention mechanism and various architecture-related networks was compared with the CNN-SA. The comparison models included CNN, CNN-CBAM [19], and CNN-LSTM [20], all of which had parallel network structures. CNN served as a baseline for comparison.…”

Section: Performance Comparisonmentioning

confidence: 99%

Multi-Sensor Data Fusion Method Based on Self-Attention Mechanism

Lin,

Chao,

Yan

et al. 2023

Applied Sciences

View full text Add to dashboard Cite

In 3D reconstruction tasks, single-sensor data fusion based on deep learning is limited by the integrity and accuracy of the data, which reduces the accuracy and reliability of the fusion results. To address this issue, this study proposes a multi-sensor data fusion method based on a self-attention mechanism. A multi-sensor data fusion model for acquiring multi-source and multi-modal data is constructed, with the core component being a convolutional neural network with self-attention (CNN-SA), which employs CNNs to process multi-source and multi-modal data by extracting their features. Additionally, it introduces an SA mechanism to weigh and sum the features of different modalities, adaptively focusing on the importance of different modal data. This enables mutual support, complementarity, and correction among the multi-modal data. Experimental results demonstrate that the accuracy of the CNN-SA network is improved by 72.6%, surpassing the improvements of 29.9% for CNN-CBAM, 23.6% for CNN, and 11.4% for CNN-LSTM, exhibiting enhanced generalization capability, accuracy, and robustness. The proposed approach will contribute to the effectiveness of multi-sensor data fusion processing.

show abstract

Improving performance of deep learning models for 3D point cloud semantic segmentation via attention mechanisms

Cited by 19 publications

References 21 publications

AFRNet: Anchor-Free Object Detection Using Roadside LiDAR in Urban Scenes

AFRNet: Anchor-Free Object Detection Using Roadside LiDAR in Urban Scenes

Efficient multiscale spatial attention 3D abdominal multiorgan segmentation model

Multi-Sensor Data Fusion Method Based on Self-Attention Mechanism

Contact Info

Product

Resources

About