Object Attribute Detection via Partial Area Voting
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional object detection methods, such as the generalized Hough transform, often result in false detections, especially when dealing with monotonous patterns like parallel lines, leading to incorrect identification of the target object's position.
Innovation Solution
The method involves detecting a partial area of the target object from an input image, evaluating the appropriateness of the detection results, and performing voting based on these evaluations to accurately identify the object's attributes, including position, orientation, and type.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If the generalized Hough transform is used to detect object position and orientation, then the detection process can be performed using shape information of the target object, but false detections occur when monotonous patterns such as parallel lines are present in the input image
Solution Approach 1:
The patent segments the target object into multiple partial areas (first partial area and second partial area) and performs detection on each segment separately. By dividing the object into distinct parts with different shapes and positions, the voting results from each segment contribute differently to the final detection, preventing false detections caused by monotonous patterns. This segmentation ensures that even if one area produces unreliable votes, other areas can provide accurate positioning information.
2Productivity
If voting is performed based on detection results from monotonous patterns, then the voting process can be completed, but all detection results become identical leading to false detection at incorrect positions
Solution Approach 1:
The patent applies local quality by making different partial areas contribute differently to the voting process based on their individual characteristics. Each partial area's voting weight and influence are determined by its specific shape, position, and detection reliability. This ensures that areas with monotonous patterns that produce identical voting results have reduced influence, while areas with distinctive features provide more reliable positioning information, thereby improving overall measurement precision.
Data Source
AI summary
An information processing method includes detecting a partial area configuring a target object from an input image, evaluating appropriateness of the detection result, voting with respect to the target object based on the detection result and the evaluation result, and identifying an attribute of the target object based on the voting result.