Method, apparatus, apparatus and medium for processing sparse depth map of lidar

Abstract

Embodiments of the present application disclose a method of processing a lidar sparse depth map, Neural network training method, vehicle intelligent control method, obstacle avoidance navigation method, device, electronic equipment, computer-readable storage medium and computer program, wherein, the processing method of lidar sparse depth map includes: inputting lidar sparse depth map to neural network; Acquiring, by the neural network, at least two feature maps of different scales of the depth map, The effective point feature fusion processing is respectively carried out for the at least twofeature maps of different scales, and the processed depth map is obtained according to the result of the effective point feature fusion processing, and the number of the effective points in the processed depth map is larger than the number of the effective points in the laser radar sparse depth map.

Description

technical field [0001] The present application relates to computer vision technology, in particular to a processing method of a laser radar sparse depth map, a processing device of a laser radar sparse depth map, a vehicle intelligent control method, a vehicle intelligent control device, an obstacle avoidance navigation method, an obstacle avoidance navigation device, A neural network training method, a neural network training device, an electronic device, a computer-readable storage medium, and a computer program. Background technique [0002] LiDAR can obtain the depth information of objects in the surrounding scene through scanning, and the depth information can form a LiDAR projection map. The value of a point in the lidar projection map usually represents the depth value of the point. A lidar projection map may also be called a lidar depth map. [0003] The lidar projection map can be used to assist in tasks such as semantic segmentation and target detection, and can ...

AI Technical Summary

Problems solved by technology

[0004] However, due to factors such as lidar hardware constraints, the lidar projecti

Images