首页>
外国专利>
Post-processing method and apparatus for detecting lanes for planning travel route of autonomous vehicle using segmentation score map and clustering map
Post-processing method and apparatus for detecting lanes for planning travel route of autonomous vehicle using segmentation score map and clustering map
展开▼
机译:利用分割得分图和聚类图检测车道以规划自动驾驶汽车行驶路线的后处理方法和装置
展开▼
页面导航
摘要
著录项
相似文献
摘要
PROBLEM TO BE SOLVED: To provide a method for smoothly performing post-processing for lane detection. A lane detection method obtains a segmentation score map and a clustering map generated by CNN, determines a lane element including pixels forming a lane by referring to the segmentation score map, and determines a lane element as a lane element. , The segmentation score map and the clustering map are used to generate seed information S03, the seed information is used to generate a base model, the lane anchor is generated S04, and the lane anchor is referenced. A lane part blob is generated S05, a lane candidate group is detected by referring to the lane blob S06, and a lane is detected by applying a line fitting calculation to the lane candidate group to generate a lane model S07. [Selection diagram] Figure 2
展开▼