Attribute Reliability Calculation Using Multi-Image Correlation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional semantic segmentation technologies face challenges in achieving sufficient accuracy and efficiency due to the high processing load required for attribute identification by removing network units multiple times for each input image, and reliability calculation is limited by using only one image.
Innovation Solution
An information processing apparatus that acquires multiple pieces of observation information, generates attribute information, and sets reliability based on the correlation of attribute information across multiple images, reducing the need for repeated attribute identification and improving calculation accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If attribute identification is conducted by removing network units multiple times for one input image, then reliability calculation is possible, but processing load increases significantly
Solution Approach 1:
The patent performs attribute identification multiple times in advance by removing different network units before final reliability calculation. This preliminary action allows the system to obtain multiple attribute identification results that can be used for reliability calculation, thereby improving reliability while maintaining processing efficiency through pre-computation.
2Reliability
If attribute identification is conducted by removing network units multiple times, then reliability can be calculated, but processing load becomes excessively large
Solution Approach 1:
The system performs attribute identification multiple times in advance by removing different network units, storing the results for later reliability calculation. This preliminary action reduces computational energy consumption during final processing by avoiding redundant calculations, thereby improving reliability while reducing energy loss.
3Productivity
If reliability is calculated using only one image, then processing is simple, but sufficient accuracy is difficult to achieve
Solution Approach 1:
The patent merges attribute identification results from multiple images to calculate reliability. By combining information from multiple images processed through the network with different units removed, the system achieves more accurate reliability calculation while maintaining reasonable processing complexity through result aggregation.
4Reliability
If attribute identification is performed multiple times with unit removal, then reliability assessment is possible, but processing time increases
Solution Approach 1:
The system performs attribute identification multiple times in advance by removing different network units and stores the results. This preliminary action enables reliable reliability calculation while reducing processing time during final operations, as the multiple identification results are already available for direct reliability computation without re-processing.
Data Source
AI summary
According to an embodiment, an information processing apparatus includes a memory having computer executable components stored therein; and a processing circuit communicatively coupled to the memory. The processing circuit acquires a plurality of pieces of observation information of surroundings of a moving body, generates a plurality of pieces of attribute information of the surroundings of the moving body on the basis of the plurality of pieces of observation information, and sets a reliability of the attribute information of the surroundings of the moving body on the basis of correlation of the plurality of pieces of attribute information.