A class-level 6d pose and size estimation method and device

Abstract

The present invention provides a class-level 6D pose and size estimation method and device, which perform 6D pose and size estimation based on the depth observation data of the target object in the image and the internal parameters of the camera that captures the image, and is characterized by comprising the following steps: Based on the camera internal parameters, the depth observation data is reprojected to three-dimensional space and normalized to obtain part of the observation point cloud data of the target object; according to the target object, the normalized template point cloud data of the same type as the target type is selected from the template point cloud database; Part of the observed point cloud data and the normalized template point cloud data use the geometric feature learning network to estimate the 6D pose and size of the target object, and obtain the corresponding predicted pose and predicted size. The class-level 6D pose and size estimation method and device of the present invention can accurately estimate the 6D pose and size of instance objects whose classes are known but not seen, and have high accuracy, generalization ability and practicability.

Description

technical field [0001] The invention belongs to the field of data recognition, and in particular relates to a category-level 6D pose and size estimation method and device. Background technique [0002] Accurate estimation of the 6D pose (3D displacement and 3D rotation in space) of a target object is particularly important in augmented reality, scene understanding tasks, and especially robotic applications. At present, most of the existing technologies have achieved very high accuracy in the "instance-level" 6D object pose estimation, but these methods rely on the accurate 3D model of the relevant target object provided in advance, which limits the generalization of the algorithm to unknown instance capability. [0003] In recent years, "class-level" 6D object pose and size estimation techniques have attempted to address the above problems. Most of the current methods consider extracting features from color (RGB) or RGB-depth (RGB-D) channels for estimation. However, due ...

AI Technical Summary

Problems solved by technology

However, there are inherent differences between the RGB and D channels, and the fusion method in [2] is difficult to fully explore and utilize the geometric information from the D channel.

[0006] In [3], the real-time 6D pose estimation of the target object is performed by means of frame tracking, but this method needs to provide the initial 6D pose and size of the target object in advance. In addition, a network model in this method can only be used for Objects within a single class, limiting its potential to expand to mobile devices

[0007] In summary, the existing category-level 6D pose and size estimation methods have poor generalization ability

Images