Multi-sensor Object Recognition Using Confidence Adjustment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In environments like retail facilities, accurate inventory management is hindered by the complexity and variability of object arrangements, lighting conditions, and the need for costly image sensors to capture entire shelf modules, leading to reduced accuracy in object identification and status information derivation.
Innovation Solution
A method and system for object detection using a mobile automation apparatus equipped with multiple image sensors capturing overlapping regions, which identifies candidate subsets of object indicators, adjusts confidence levels, and generates single output indicators for each cluster, improving accuracy by handling partial and duplicated object depictions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple image sensors are used to capture overlapping regions, then object identification accuracy is improved, but device complexity increases
Solution Approach 1:
The system divides the imaging task into multiple segments by using several image sensors to capture different overlapping regions of the same scene. Each sensor captures a portion of the object, and the system processes these segmented images to reconstruct complete object indicators, thereby improving identification accuracy while managing device complexity through distributed sensing.
Solution Approach 2:
The system merges multiple partial object indicators from different image sensors into a single comprehensive object indicator. By combining the detected features, bounding boxes, and confidence levels from multiple sensors, the system achieves more accurate and complete object identification than any single sensor could provide alone.
2Loss of information
If image sensors capture entire shelf modules, then complete object information is obtained, but cost increases
Solution Approach 1:
Instead of using expensive sensors to capture the entire shelf module, the system uses multiple less expensive image sensors to capture only the necessary overlapping regions containing objects. This partial action approach obtains complete object information without the excessive cost of capturing entire shelf modules, achieving cost-effective inventory management.
3Measurement precision
If overlapping images are integrated, then accuracy in complex environments is improved, but processing time increases
Solution Approach 1:
The system performs preliminary actions by detecting candidate object indicators in each image before full integration. By pre-identifying potential objects, filtering candidates based on confidence levels, and preparing bounding boxes in advance, the system reduces the computational burden during final integration, thereby managing processing time while maintaining high accuracy in complex environments.
Data Source
AI summary
A method of object detection includes obtaining a set of images depicting overlapping regions of an area containing a plurality of objects. Each image includes input object indicators defined by input bounding boxes, input confidence level values, and object identifiers. The method includes identifying candidate subsets of input object indicators in adjacent images. Each candidate subset has input overlapping bounding boxes in a common frame of reference, and a common object identifier. The method includes adjusting the input confidence levels for each input object indicator in the candidate subsets; selecting clusters of the input object indicators satisfying a minimum input confidence threshold, having a common object identifier, and having a degree of overlap satisfying a predefined threshold; and detecting an object by generating a single output object indicator for each cluster, the output object indicator having an output bounding box, an output confidence level value, and the common object identifier.