6D pose recognition method and device and computer storage medium

Abstract

The invention discloses a 6D pose recognition method and device and a computer storage medium, and relates to the technical field of machine vision. The method comprises the following steps: acquiringa first target depth map, a first target RGB map and first segmentation information of a to-be-measured object; wherein the first segmentation information comprises a first segmentation mask; obtaining first point cloud information corresponding to the first target depth map; inputting the first point cloud information, the first target RGB image and the first segmentation information into a trained pose recognition network; and acquiring 6D pose information of the to-be-measured object according to an output result of the pose recognition network. The features of the first target depth map,the first target RGB map and the first segmentation information of the to-be-measured object are extracted through the trained pose recognition network to obtain the 6D pose information, the featuresof the to-be-measured object can be fully utilized, and the 6D pose can be accurately measured under the complex environments such as strong reflection, weak texture and serious shielding. Therefore,the accuracy of pose recognition is improved.

Description

technical field [0001] The present invention relates to the field of artificial intelligence, in particular to a method, device and computer storage medium for 6D pose recognition. Background technique [0002] With the popularity of depth cameras and the development of 3D vision research, robotic arms have begun to undertake complex tasks such as intelligent sorting and flexible loading and unloading. Taking the grasping of parts in an industrial production line as an example, accurate measurement of the 6D pose of the target part is the basis for guiding the robot to complete the grasping operation. Among them, pose is a relative concept, referring to the distance between two coordinate systems. The displacement and rotation transformation, and the two coordinate systems represent the coordinate system established by the initial position on the object and the position after rotation and translation, 6D refers to 6 degrees of freedom, including translation of 3 degrees of f...

AI Technical Summary

Problems solved by technology

However, industrial parts have the characteristics of complex structure, single surface texture, strong reflection, etc., and they are seriously blocked by each other; through the method of 6D pose recognition based on point cloud information, due to the inherent sparseness and redundancy of point cloud data and disorder, requires a large amount of data calculation to ensure the accuracy of pose recognition; while the method of recognition based on RGB images, in the measurement scene of low-textured objects, the accuracy is low

Images