The efficacy of autonomous driving systems is fundamentally linked to their ability to accurately perceive and interpret their surroundings. The investigation addresses the challenges inherent in the sparse and unordered 3D point cloud data produced by LiDAR sensors by introducing a refined PointPillars network architecture. This innovative framework, which seamlessly integrates Transformer modules and ECA-PP (Efficient Channel Attention PointPillars) modules, substantially elevates the precision and efficiency of 3D object detection tasks. The Transformer module adeptly harnesses self-attention mechanisms to extract and process global contextual information from point cloud data, capturing not only local features but also comprehending the overall structure and layout of the scene. Meanwhile, the ECA-PP module enhances feature discrimination by focusing on channel attention, effectively distinguishing and emphasizing salient features crucial for object detection. Comprehensive experimental evaluations, conducted using the KITTI dataset and nuScenes dataset, have yielded compelling evidence that the proposed algorithm significantly outperforms current technologies in terms of detection accuracy for various categories, including cars, pedestrians, and cyclists.
Montiel-MarínSLlamazaresAAntunesM, et al. Point cloud painting for 3D object detection with camera and automotive 3+1D RADAR fusion. Sensors2024; 24: 20.
2.
GuoLLuKHuangL, et al. Pillar-based multilayer pseudo-image 3D object detection. J Electron Imaging2024; 33: 14.
3.
MaoJGShiSSWangXG, et al. 3D object detection for autonomous driving: a comprehensive survey. Int J Comput Vis2023; 131: 1909–1963.
4.
XieCLinCYZhengXY, et al. Dense sequential fusion: point cloud enhancement using foreground mask guidance for multimodal 3-D object detection. IEEE Trans Instrum Meas2024; 73: 15.
5.
LiHSLuYL. 3D object detection based on point cloud in automatic driving scene. Multimed Tools Appl2024; 83: 13029–13044.
6.
LangAHVoraSCaesarH, et al. PointPillars: fast encoders for object detection from point clouds. In: IEEE/CVF conference on computer vision and pattern recognition (CVPR), Long Beach, CA, USA, 15–20 June 2019, pp.12689–12697. New York: IEEE.
7.
GeigerALenzPUrtasunR. Are we ready for autonomous driving? The KITTI vision benchmark suite. In: IEEE conference on computer vision and pattern recognition (CVPR), Providence, RI, USA, 16–21 June 2012, pp.3354–3361. New York: IEEE.
8.
CaesarHBankitiVLangHA, et al. nuScenes: a multimodal dataset for autonomous driving. In: 2020 IEEE/CVF conference on computer vision and pattern recognition (CVPR), Seattle, WA, USA, 13–19 June 2020.
9.
QiCRSuHMoKC, et al. PointNet: deep learning on point sets for 3D classification and segmentation. In: 30th IEEE/CVF conference on computer vision and pattern recognition (CVPR), Honolulu, HI, USA, 21–26 July 2017, pp.77–85. New York: IEEE.
10.
ShiSSWangXGLiHS, et al. PointRCNN: 3D object proposal generation and detection from point cloud. In: 32nd IEEE/CVF conference on computer vision and pattern recognition (CVPR), Long Beach, CA, USA, 15–20 June 2019, pp.770–779. New York: IEEE.
11.
ZhouYTuzelO. VoxelNet: end-to-end learning for point cloud based 3D object detection. In: 31st IEEE/CVF conference on computer vision and pattern recognition (CVPR), Salt Lake City, UT, USA, 18–23 June 2018, pp.4490–4499. New York: IEEE.
GuoMHCaiJXLiuZN, et al. PCT: point cloud transformer. Comput Vis Media2021; 7: 187–199.
14.
LiuZJYangXYTangHT, et al. FlatFormer: flattened window attention for efficient point cloud transformer. In: IEEE/CVF conference on computer vision and pattern recognition (CVPR), Vancouver, BC, Canada, 18–22 June 2023, pp.1200–1211. New York: IEEE.
15.
WangXKJiaXZhangMY, et al. Object detection in 3D point cloud based on ECA mechanism. J Circuits Syst Comput2023; 32: 17.
16.
VaswaniAShazeerNParmarN, et al. Attention is all you need. In: 31st annual conference on neural information processing systems (NIPS), Long Beach, CA, USA, 4–9 December 2017.
17.
ZhaoHSJiangLJiaJY, et al. Point transformer. In: 18th IEEE/CVF international conference on computer vision (ICCV), 2021, pp.16239–16248. New York: IEEE.
18.
XiaYXuYSLiS, et al. SOE-Net: a self-attention and orientation encoding network for point cloud based place recognition. In: IEEE/CVF conference on computer vision and pattern recognition (CVPR), Nashville, TN, USA, 20–25 June 2021, pp.11343–11352. New York: IEEE.
19.
HeCHLiRHLiS, et al. Voxel set transformer: a set-to-set approach to 3D object detection from point clouds. In: IEEE/CVF conference on computer vision and pattern recognition (CVPR), New Orleans, LA, USA, 2022, pp.8407–8417. New York: IEEE.
20.
ZhangLMengHYanYB, et al. Transformer-based global PointPillars 3D object detection method. Electronics2023; 12: 13.
21.
CholletF. Xception: deep learning with depthwise separable convolutions. In: 30th IEEE/CVF conference on computer vision and pattern recognition (CVPR), Honolulu, HI, USA, 21–26 July 2017, pp.1800–1807. New York: IEEE.
22.
WangQWuBZhuP, et al. ECA-Net: efficient channel attention for deep convolutional neural networks. arXiv preprint arXiv:1910.03151v1, 2020.
23.
HuJShenLAlbanieS, et al. Squeeze-and-excitation networks. IEEE Trans Pattern Anal Mach Intell2020; 42: 2011–2023.
24.
WangSQChenM. A LiDAR multi-object detection algorithm for autonomous driving. Appl Sci2023; 13: 16.
25.
LinTYGoyalPGirshickR, et al. Focal loss for dense object detection. IEEE Trans Pattern Anal Mach Intell2020; 42: 318–327.
26.
ChenXZMaHMWanJ, et al. Multi-view 3D object detection network for autonomous driving. In: 30th IEEE/CVF conference on computer vision and pattern recognition (CVPR). Honolulu, HI, USA, 21–26 July 2017, pp.6526–6534. New York: IEEE.
27.
EveringhamMVan GoolLWilliamsCKI, et al. The Pascal Visual Object Classes (VOC) challenge. Int J Comput Vis2010; 88: 303–338.
28.
QiCRLiuWWuCX, et al. Frustum PointNets for 3D object detection from RGB-D data. In: 31st IEEE/CVF conference on computer vision and pattern recognition (CVPR), Salt Lake City, UT, USA, 18–23 June 2018, pp.918–927. New York: IEEE.
29.
SimonMMilzSAmendeK, et al. Complex-YOLO: an Euler-region-proposal for real-time 3D object detection on point clouds. In: 15th European conference on computer vision (ECCV), Munich, Germany, 8–14 September 2018, pp.197–209. Cham, Switzerland: Springer Cham.
30.
LiuZYeXTanX, et al. Stereodistill: pick the cream from lidar for distilling stereo-based 3d object detection. Proc AAAI Conf Artif Intell2023; 37: 1790–1798.
31.
HayeonOYangCHuhK. SeSame: simple, easy 3D object detection with point-wise semantics. In: ChoMLaptevITranD, et al. (eds) Asian conference on computer vision. Singapore: Springer, 2025, pp.211–227.
32.
MaoJXueYNiuM, et al. Voxel transformer for 3D object detection. In: 18th IEEE/CVF international conference on computer vision (ICCV), 2021, pp.3144–3153. New York: IEEE.
33.
LiYChenYQiX, et al. Unifying voxel-based representation with transformer for 3d object detection. Adv Neural Inf Process Syst2022; 35: 18442–18455.
34.
JiangYZhangLMiaoZ, et al. PolarFormer: multi-camera 3D object detection with polar transformer. Proc AAAI Conf Artif Intell2023; 37: 1042–1050.
35.
LiZWangWLiH, et al. BEVFormer: learning Bird’s-Eye-View representation from multi-camera images via spatiotemporal transformers. arXiv preprint arXiv:2203. 17270, 2022.
