Target Pose Detection Using Segmented Pixel Relationship Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing target detection methods suffer from low detection efficiency, particularly in scenarios involving industrial production where intelligent robots are used for workpiece assembly.
Innovation Solution
A target detection method that includes instance segmentation to obtain a segmentation mask, followed by matching position relationship features between target and standard pixels to determine pose information, utilizing a pre-trained instance segmentation model with a feature extraction network and feature fusion to enhance accuracy and efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional target detection methods are used, then the detection process is simple, but the detection efficiency is low
Solution Approach 1:
The patent applies segmentation by dividing the target object into multiple keypoint regions and extracting features from each region separately. The instance segmentation model segments the target image to identify distinct keypoint locations, and then position relationship features are extracted from these segmented regions. This segmentation approach reduces the amount of data that needs to be matched while maintaining high detection efficiency.
Solution Approach 2:
The patent extracts only the essential position relationship features from the target object rather than processing the entire image or all image features. By using the segmentation mask to identify keypoint regions and extracting position relationship features specifically from these regions, the method extracts only the necessary information needed for detection, thereby improving efficiency without requiring complex processing of the full image.
2Measurement precision
If position relationship features between all pixels are matched, then comprehensive pose information can be obtained, but the matching data volume is large and processing is slow
Solution Approach 1:
The patent extracts only the essential position relationship features from keypoint regions identified through instance segmentation, rather than processing all image pixels. The segmentation mask is used to locate keypoint regions, and position relationship features are extracted specifically from these regions, significantly reducing the data volume that needs to be matched while maintaining the accuracy needed for pose information retrieval.
Solution Approach 2:
The patent applies local quality by focusing computational resources on specific keypoint regions of the target object rather than uniformly processing the entire image. The instance segmentation model identifies which regions contain keypoints, and the position relationship feature extraction is concentrated on these localized regions. This localized approach reduces overall processing time while maintaining the precision needed for accurate pose estimation.
Data Source
AI summary
Embodiments of this application provide a target detection method and apparatus, an electronic device, and a computer storage medium. The method includes: obtaining a target image including a target object; performing instance segmentation on the target image, to obtain a segmentation mask corresponding to the target object; obtaining, based on the segmentation mask, position relationship features between target pixels in a target region in which the target object is located in the target image; obtaining position relationship features between standard pixels in a preset region of interest in a standard image, where the standard image includes a standard object corresponding to the target object; and matching the position relationship features between the target pixels and the position relationship features between the standard pixels, to obtain a correspondence between the target pixels and the standard pixels, and obtaining pose information of the target object based on the correspondence.