The application discloses a kind of cone
barrel racecourse boundary classification method
system, device and storage medium, wherein method includes: obtaining the cone
barrel detection result based on
laser point cloud;Wherein, the spatial coordinates of each cone
barrel are cone barrel detection result;Cone barrel detection result is provided as input to pre-established classification model, and the
spatial distribution interrelationship feature and intermediate layer feature of cone barrel are extracted;The
spatial distribution interrelationship feature and intermediate feature of cone barrel are aggregated by the classification model, and the
classification result is obtained according to the aggregation result;Wherein,
classification result is used to indicate the boundary category of each cone barrel.The application can obtain more reliable, more accurate, more quickly cone barrel racecourse boundary information compared with the method of image and
laser point cloud fusion in prior art for racecourse cone barrel detection, to solve the problem of unreliable,
high complexity in the cone barrel racecourse boundary
classification scheme of prior art.The application can be widely applied in the field of unmanned driving technology.