Image recognition method, vehicle control method, and electronic device
By fusing point clouds from multiple image frames into a voxel space, constructing voxel structures and extracting features, and then using feature masks and machine learning models for image recognition, the problem of low efficiency and accuracy of image recognition in automated delivery vehicles is solved, thus improving driving safety.
CN120877265BActive Publication Date: 2026-07-24BEIJING SANKUAI ONLINE TECH CO LTD +1
View PDF 2 Cites 0 Cited by
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- BEIJING SANKUAI ONLINE TECH CO LTD
- Filing Date
- 2024-04-30
- Publication Date
- 2026-07-24
AI Technical Summary
Technical Problem
The low efficiency and accuracy of image recognition computing in automated delivery vehicles lead to lower driving safety.
Method used
Point clouds from multiple frames of images are fused into voxel space to construct voxel structures. Voxel features are extracted through attribute representations on feature maps. Target point clouds are determined using feature masks, and image recognition is performed using machine learning models.
Benefits of technology
This improves the accuracy and efficiency of image recognition, enhancing the driving safety of automated delivery vehicles.
✦ Generated by Eureka AI based on patent content.
Smart Images
Figure CN120877265B_ABST
Abstract
The present disclosure relates to an image recognition method, a vehicle control method and an electronic device, comprising: fusing point clouds in multiple frames of images into voxel space, and constructing a voxel structure on a single frame representation; extracting attribute features of voxels in each voxel grid in the voxel structure on the attribute representation according to the attribute representation on the feature map; determining feature masks of voxels corresponding to different coordinate axes under a window structure according to different types of voxel coordinates and voxel indexes of voxels in the voxel structure, wherein the different coordinate axes are determined according to the voxels in the bird's eye view; determining voxels with different indexes in the feature masks as target point clouds according to the point clouds corresponding to the multiple frames of images; and performing image recognition on the multiple frames of images according to the attribute features corresponding to the target point clouds to obtain an image recognition result. The scale of the feature map on the coordinate axis can be maintained, the accuracy of image recognition can be improved, and the efficiency of image recognition can be improved by introducing a window to determine the feature mask.
Need to check novelty before this filing date? Find Prior Art