Method and device for road image segmentation based on point cloud data and electronic equipment

Abstract

The embodiment of the invention discloses a method and device for road image segmentation based on point cloud data and electronic equipment. The method comprises the steps of obtaining point cloud data of a road image, and performing dimensionality reduction to obtain corresponding three-dimensional tensor data; and segmenting the road image according to the three-dimensional tensor data by usinga lightweight deep neural network, wherein the lightweight deep neural network is obtained by training based on the three-dimensional tensor data. According to the invention, dimensionality reductionprocessing is performed on the collected point cloud data by adopting a spherical coordinate system, so that a sparse problem of the original collected point cloud data can be solved, and the speed of performing operation by using the deep neural network in a later period can be improved. Furthermore, the lightweight unet deep neural network framework based on a mobilenet network structure is utilized, thereby being capable of operating based on embedded equipment such as an FPGA, being capable of effectively improving the operation speed, and reducing the operation power consumption of the equipment.

Description

technical field [0001] This description relates to the field of computer technology, in particular to a method, device and electronic equipment for road image segmentation based on point cloud data. Background technique [0002] In the field of autonomous driving, the importance of road segmentation lies in its ability to provide reliable traffic road areas for autonomous vehicles. The traditional road segmentation method is often based on the image data collected by the camera; however, due to the lack of 3D information of the road, the accuracy is lacking. [0003] In order to solve the problem of the accuracy of road segmentation, lidar has been used more and more recently. Although it is expensive, it can provide 3D information and can also achieve precise positioning. In recent years, the fusion of lidar and image data has been used to further improve the accuracy. But their common problem is that the amount of data to be processed is large, which will lead to slow da...

AI Technical Summary

Problems solved by technology

But their common problem is that the amount of data processed is large, which will lead to slow data processing

Images