3D Pose Estimation for Symmetric Objects via Mirror Image Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current systems face challenges in accurately determining the pose of three-dimensional symmetric or nearly-symmetric objects from images due to the similarity of multiple orientations, which complicates the processing and estimation of their correct pose.
Innovation Solution
The method involves obtaining an image or image patch, determining if the rotation angle is within a predetermined range, creating a mirror image, and using a trained regressor to estimate the three-dimensional pose based on two-dimensional projections of a bounding box, thereby reducing processing time and improving accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional pose estimation methods are used for symmetric objects, then multiple similar orientations cannot be distinguished, but using mirror image transformation and classification improves orientation discrimination accuracy
Solution Approach 1:
The method segments the pose estimation problem by first classifying the rotation angle into discrete ranges (0-60°, 60-120°, etc.), then applying specific transformations (mirror image or rotation) for each range. This segmentation allows the system to handle symmetric objects by breaking down the continuous rotation space into manageable segments with known transformation rules.
Solution Approach 2:
The method performs preliminary actions by determining the rotation angle range and applying appropriate transformations (mirror image or rotation) before the final pose estimation. This preliminary classification and transformation step prepares the image data in a standardized form that makes subsequent pose estimation accurate and unambiguous.
2Loss of time
If comprehensive pose estimation is performed without preliminary classification, then all orientations can be estimated, but processing time increases significantly
Solution Approach 1:
The method segments the pose estimation problem by first classifying the rotation angle into discrete ranges (0-60°, 60-120°, etc.), then applying specific transformations (mirror image or rotation) for each range. This segmentation allows the system to handle symmetric objects by breaking down the continuous rotation space into manageable segments with known transformation rules.
Solution Approach 2:
The method performs preliminary actions by determining the rotation angle range and applying appropriate transformations (mirror image or rotation) before the final pose estimation. This preliminary classification and transformation step prepares the image data in a standardized form that makes subsequent pose estimation accurate and unambiguous.
Data Source
AI summary
The present disclosure describes methods, apparatuses, and non-transitory computer-readable mediums for estimating a three-dimensional (“3D”) pose of an object from a two-dimensional (“2D”) input image which contains the object. Particularly, certain aspects of the disclosure are concerned with 3D pose estimation of a symmetric or nearly-symmetric object. An image or a patch of an image includes the object. A classifier is used to determine whether a rotation angle of the object in the image or the patch of the image is within a first predetermined range. In response to a determination that the rotation angle is within the first predetermined range, a mirror image of the object is determined. Two-dimensional (2D) projections of a three-dimensional (3D) bounding box of the object are determined by applying a trained regressor to the mirror image of the object in the image or the patch of the image. The 3D pose of the object is estimated based on the 2D projections.