目标检测方法、装置、终端设备及计算机可读存储介质
By dividing radar point cloud data into concentrated and sparse target regions and using voxel partitioning methods of different sizes, combined with three-dimensional and two-dimensional networks to extract feature information, the problems of low detection accuracy and low efficiency in existing technologies are solved, achieving higher detection accuracy and efficiency.
CN115861626BActive Publication Date: 2026-07-17WUHAN WANJI INFORMATION TECH
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- WUHAN WANJI INFORMATION TECH
- Filing Date
- 2022-10-25
- Publication Date
- 2026-07-17
AI Technical Summary
Technical Problem
Existing target detection methods cannot fully extract feature information of concentrated target areas within the radar detection range, resulting in low detection accuracy. Furthermore, improving feature extraction accuracy involves large amounts of data processing and low efficiency.
Method used
Point cloud data is divided into concentrated and sparse regions within the height range, and different voxel sizes are used for the division. Smaller voxel sizes are used for concentrated regions, while larger voxel sizes are used for sparse regions. Feature information is extracted by combining 3D and 2D networks.
Benefits of technology
It improves target detection accuracy, reduces data processing volume, and enhances detection efficiency.
✦ Generated by Eureka AI based on patent content.
Smart Images
Figure CN115861626B_ABST
Abstract
本申请适用于检测技术领域，提供了一种目标检测方法、装置、终端设备及计算机可读存储介质，包括：将点云数据对应的高度范围划分为第一范围和第二范围，其中，所述第一范围表示目标集中区域，所述第二范围表示目标稀疏区域，所述点云数据通过雷达获取；根据第一尺寸对所述第一范围内的点云数据进行体素划分，得到第一体素集合；根据第二尺寸对所述第二范围内的点云数据进行体素划分，得到第二体素集合，其中，所述第二尺寸大于所述第一尺寸；根据所述第一体素集合和所述第二体素集合检测目标对象。通过上述方法，可以有效提高目标检测精度。
Need to check novelty before this filing date? Find Prior Art