Multi-target object identification and positioning method and device, electronic equipment, storage medium

Abstract

The present invention provides a multi-target identification and positioning method and device, electronic equipment, and a storage medium. The method includes: collecting a first image of an object to be detected, and using a preset feature extraction network to perform feature extraction on the first image. Point extraction, obtaining the reference feature point template of the object to be detected; collecting the second image, dividing the second image into a plurality of second sub-images, and using the preset feature extraction network to perform feature points on the plurality of second sub-images respectively Extract, and compare the similarity between the extracted feature and the reference feature point template, and determine the extracted feature point whose similarity reaches the set threshold as the candidate target feature point; calculate the distance between the candidate target feature point and the corresponding reference feature point template The transformation matrix of the object to be detected in the template is used to determine the reference coordinates of the object to be detected in the second image based on the transformation matrix, and provide it to the operating equipment. The invention recognizes the detection object based on the 2D image, which is accurate and improves the calculation efficiency.

Description

technical field [0001] The invention relates to target object recognition technology, in particular to a multi-target object recognition and positioning method and device, electronic equipment, and a storage medium. Background technique [0002] In recent years, the sorting of small package items on the assembly line is mainly based on the identification and positioning of the object feature points in the image, and the identification and positioning results are input into the robotic arm, so as to realize automatic identification and grasping, reduce labor costs, and improve production efficiency. At present, it is mainly to use deep learning detection and classification technology to realize multi-target positioning, select the optimal grasping candidate target, and provide the position information to the manipulator for grasping. At present, the mainstream algorithm is still based on 3D, using deep learning to process point cloud information, obtain segmentation targets, ...

AI Technical Summary

Problems solved by technology

This leads to a huge amount of calculations and a large number of iterations, resulting in low recognition and positioning efficiency and excessive computing resources.

Images