关键词: 激光雷达, 目标检测, 点云掩码, 不确定性估计

Abstract: The 3D point cloud data acquired by LiDAR is crucial for autonomous driving. However, the annotation of point cloud data is difficult; the available data is limited; and there is usually a large degree of uncertainty in its labels, which constrains the effectiveness of deep learning-based 3D object detection models. To address these issues, this paper proposes a point cloud masking strategy to construct a pre-training dataset, combined with transfer learning to improve detection accuracy. Additionally, a Gaussian distribution-based localization uncertainty estimation modeling method is proposed, enabling the object detection model to predict the localization uncertainty of each coordinate while predicting the bounding box coordinates. Experimental results demonstrate that, without significantly increasing algorithmic complexity, the proposed method effectively reduces false detections and significantly improves the accuracy of object detection.

Key words: LiDAR, object detection, point cloud mask, uncertainty estimation