基于稀疏自注意力图神经网络的三维目标检测

doi:10.3778/j.issn.1002-8331.2309-0219

摘要/Abstract

摘要： 三维目标检测是自动驾驶环境感知中最重要的技术之一。为了解决远距离漏检问题，提升三维目标检测的效果，提出一种基于稀疏自注意力图神经网络的三维目标检测方法（SSA-GNN），在采样关键点阶段，提出动态区域并行采样法，通过采样区域过滤，场景划分为扇区，融合动态最远体素采样的方式，以保持关键点均匀分布、加速采样同时提升前景点比例。在细化建议框阶段，利用图神经网络在点之间建立联系，通过迭代的消息传递来更好地建模上下文信息和聚合领域信息，并改进多头自注意机制来更好地关注特征聚合后领域中的重要关系，从而提高算法检测性能。SSA-GNN在KITTI公开数据集上进行测试，与基线网络PointPillars、SECOND和PointRCNN相比，在困难等级指标下，Car类平均精度分别提升了7.95、5.50、6.94个百分点，结果表明SSA-GNN可有效提升三维目标检测性能。

关键词: 三维目标检测, 关键点采样, 图神经网络, 稀疏自注意力

Abstract: 3D object detection is one of the most important technologies in autonomous driving environment perception. To address the issue of long-range missed detections and enhance the effectiveness of 3D object detection, a 3D object detection method based on a sparse self-attention graph neural network (SSA-GNN) is proposed. In the stage of sampling key points, a dynamic area parallel sampling method is proposed. Through sampling area filtering, the scene is divided into sectors, and the way of dynamic farthest voxel sampling is integrated to maintain the uniform distribution of key points, accelerate sampling, and increase the proportion of foreground points. In the stage of refining the suggestion box, the graph neural network is used to establish connections between points, and the context information is better modeled and the domain information is aggregated through iterative message passing. Moreover, the multi-head self-attention mechanism is improved to pay more attention to the important relationships in the domain after feature aggregation, so as to improve the detection performance of the algorithm. SSA-GNN is tested on the KITTI public data set. Compared with the baseline network PointPillars, SECOND, and PointRCNN, the average accuracy of the Car category is increased by 7.95, 5.50, and 6.94?percentage points respectively under the difficulty level index. The results show that SSA-GNN can effectively improve the performance of 3D object detection.

Key words: 3D object detection, key point sampling, graph neural network, sparse self-attention

彭志辰, 封岸松, 王天柱, 邵鑫喆, 库涛. 基于稀疏自注意力图神经网络的三维目标检测[J]. 计算机工程与应用, 2025, 61(3): 295-305.

PENG Zhichen, FENG Ansong, WANG Tianzhu, SHAO Xinzhe, KU Tao. 3D Object Detection Based on Sparse Self-Attention Graph Neural Network[J]. Computer Engineering and Applications, 2025, 61(3): 295-305.

参考文献

[1] 李宇杰, 李煊鹏, 张为公. 基于视觉的三维目标检测算法研究综述[J]. 计算机工程与应用, 2020, 56(1): 11-24.
LI Y J, LI X P, ZHANG W G. Survey on vision-based 3D object detection methods[J]. Computer Engineering and Applications, 2020, 56(1): 11-24.
[2] 肖雨晴, 杨慧敏. 目标检测算法在交通场景中应用综述[J]. 计算机工程与应用, 2021, 57(6): 30-41.
XIAO Y Q, YANG H M. Research on application of object detection algorithm in traffic scene[J]. Computer Engineering and Applications, 2021, 57(6): 30-41.
[3] 周燕, 蒲磊, 林良熙, 等. 激光点云的三维目标检测研究进展[J]. 计算机科学与探索, 2022, 16(12): 2695-2717.
ZHOU Y, PU L, LIN L X et al. Research progress on 3D object detection of LiDAR point cloud[J]. Journal of Frontiers of Computer Science and Technology, 2022, 16(12): 2695-2717.
[4] 陆慧敏, 杨朔. 基于深度神经网络的自动驾驶场景三维目标检测算法[J]. 北京工业大学学报, 2022, 48(6): 589-597.
LU H M, YANG S. Three-dimensional object detection algorithm based on deep neural networks for automatic driving[J]. Journal of Beijing University of Technology, 2022, 48(6): 589-597.
[5] ARNOLD E, AL-JARRAH O Y, DIANATI M, et al. A survey on 3D object detection methods for autonomous driving applications[J]. IEEE Transactions on Intelligent Transportation Systems, 2019, 20(10): 3782-3795.
[6] QI C R, SU H, MO K, et al. Pointnet: deep learning on point sets for 3D classification and segmentation[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2017: 652-660.
[7] QI C R, YI L, SU H, et al. Pointnet++: deep hierarchical feature learning on point sets in a metric space[C]//Advances in Neural Information Processing Systems, 2017.
[8] SHI W, RAJKUMAR R. Point-GNN: graph neural network for 3D object detection in a point cloud[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2020: 1711-1719.
[9] KUANG H, WANG B, AN J, et al. Voxel-FPN: multi-scale voxel feature aggregation for 3D object detection from LIDAR point clouds[J]. Sensors, 2020, 20(3): 704.
[10] HE C, ZENG H, HUANG J, et al. Structure aware single-stage 3D object detection from point cloud[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2020: 11873-11882.
[11] ZHOU Y, TUZEL O. VoxelNet: end-to-end learning for point cloud based 3D object detection[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2018: 4490-4499.
[12] WANG L, SONG Z, ZHANG X, et al. SAT-GCN: self-attention graph convolutional network-based 3D object detection for autonomous driving[J]. Knowledge-Based Systems, 2023, 259: 110080.
[13] 许鑫冉, 王腾宇, 鲁才. 图神经网络在知识图谱构建与应用中的研究进展[J]. 计算机科学与探索, 2023, 17(10): 2278-2299.
XU X R, WANG T Y, LU C. Research progress of graph neural network in knowledge graph construction and application[J]. Journal of Frontiers of Computer Science and Technology, 2023, 17(10): 2278-2299.
[14] RUSCH T K, BRONSTEIN M M, MISHRA S. A survey on oversmoothing in graph neural network[J]. arXiv:2303. 10993, 2023.
[15] SHI S, JIANG L, DENG J, et al. PV-RCNN++: point-voxel feature set abstraction with local vector representation for 3D object detection[J]. International Journal of Computer Vision, 2023, 131(2): 531-551.
[16] 车运龙, 袁亮, 孙丽慧. 基于强语义关键点采样的三维目标检测方法[J]. 计算机工程与应用, 2024, 60(9): 254-260.
CHE Y L, YUAN L, SUN L H. 3D object detection based on strong semantic key point sampling[J]. Computer Engineering and Applications, 2024, 60(9): 254-260.
[17] VASWANI A, SHAZEER N, PARMAR N, et al. Attention is all you need[C]//Advances in Neural Information Processing Systems, 2017.
[18] GEIGER A, LENZ P, STILLER C, et al. Vision meets robotics: the kitti dataset[J]. The International Journal of Robotics Research, 2013, 32(11): 1231-1237.
[19] YANG H, LIU Z, WU X, et al. Graph R-CNN: towards accurate 3D object detection with semantic-decorated local graph[C]//Proceedings of the European Conference on Computer Vision, 2022: 662-679.
[20] WANG Y, SUN Y, LIU Z, et al. Dynamic graph CNN for learning on point clouds[J]. ACM Transactions on Graphics (TOG), 2019, 38(5): 1-12.
[21] 张婷, 张兴忠, 王慧民, 等. 基于图神经网络的变电站场景三维目标检测[J]. 计算机工程与应用, 2023, 59(9): 329-336.
ZHANG T, ZHANG X Z, WANG H M, et al. 3D object detection in substation scene based on graph neural network[J]. Computer Engineering and Applications, 2023, 59(9): 329-336.
[22] 李克文, 柯翠虹, 张敏, 等. 增强局部注意力的时间序列分类方法[J]. 计算机工程与应用, 2024, 60(1): 189-197.
LI K W, KE C H, ZHANG M, et al. Time series classification method with local attention enhancement[J]. Computer Engineering and Applications, 2024, 60(1): 189-197.
[23] MISRA I, GIRDHAR R, JOULIN A. An end-to-end transformer model for 3D object detection[C]//Proceedings of the IEEE/CVF International Conference on Computer Vision, 2021: 2906-2917.
[24] LAI X, LIU J, JIANG L, et al. Stratified transformer for 3D point cloud segmentation[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2022: 8500-8509.
[25] LIN T Y, GOYAL P, GIRSHICK R, et al. Focal loss for dense object detection[C]//Proceedings of the IEEE International Conference on Computer Vision, 2017: 2980-2988.
[26] GIRSHICK R. Fast R-CNN[C]//Proceedings of the IEEE International Conference on Computer Vision, 2015: 1440-1448.
[27] BELKIN M, NIYOGI P. Semi-supervised learning on riemannian manifolds[J]. Machine Learning, 2004, 56: 209-239.
[28] MA X, OUYANG W, SIMONELLI A, et al. 3D object detection from images for autonomous driving: a survey[J]. arXiv:2202.02980, 2022.
[29] TEAM O D. OpenPCDet: an open-source toolbox for 3D object detection from point clouds[EB/OL]. (2020). https://github.com/open-mmlab/OpenPCDet.
[30] LANG A H, VORA S, CAESAR H, et al. Pointpillars: fast encoders for object detection from point clouds[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2019: 12697-12705.
[31] YAN Y, MAO Y, LI B. SECOND: sparsely embedded convolutional detection[J]. Sensors, 2018, 18(10): 3337.
[32] SHI S, WANG X, LI H. PointRCNN: 3D object proposal generation and detection from point cloud[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2019: 770-779.
[33] YANG Z, SUN Y, LIU S, et al. 3DSSD: point-based 3D single stage object detector[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2020: 11040-11048.
[34] ZHANG Y, HU Q, XU G, et al. Not all points are equal: learning highly efficient point-based detectors for 3D lidar point clouds[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2022: 18953-18962.
[35] SHI S, GUO C, JIANG L, et al. PV-RCNN: point-voxel feature set abstraction for 3D object detection[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2020: 10529-10538.
[36] HE Q, WANG Z, ZENG H, et al. Stereo RGB and deeper LIDAR-based network for 3D object detection in autonomous driving[J]. IEEE Transactions on Intelligent Transportation Systems, 2022, 24(1): 152-162.
[37] PANG S, MORRIS D, RADHA H. Fast-CLOCs: fast camera-LiDAR object candidates fusion for 3D object detection[C]//Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision (WACV), 2022: 187-196.
[38] LI X, SHI B, HOU Y, et al. Homogeneous multi-modal feature fusion and interaction for 3D object detection[C]//Proceedings of the European Conference on Computer Vision. Cham: Springer Nature Switzerland, 2022: 691-707.
[39] RUSCH T K, CHAMBERLAIN B, ROWBOTTOM J, et al. Graph-coupled oscillator networks[C]//Proceedings of the International Conference on Machine Learning, 2022: 18888-18909.