Occlusion Correction Infilling for Accurate Object Attribute Measurement
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Accurate property measurements are compromised when objects of interest are occluded, leading to inaccurate surface area and attribute value assessments.
Innovation Solution
A method and system for occlusion correction that includes determining if an object of interest is occluded, removing occluding objects, infilling the occluded portions, and determining attributes using generative models and template-based approaches to enhance accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If occlusion correction is performed using generative models and template-based approaches, then measurement precision is improved, but device complexity increases
Solution Approach 1:
The occlusion correction process is divided into distinct sequential steps: detecting occluded regions, removing occluding objects, infilling occluded portions using generative models, and determining attributes. This segmentation allows each step to be optimized independently while working towards the overall goal of improving measurement precision without excessive complexity
Solution Approach 2:
The patent introduces an infilling model as an intermediary component that bridges the gap between the original measurement with occlusions and the corrected measurement. This intermediary infilling mechanism uses generative models and template-based approaches to fill occluded regions, enabling precise measurements while managing system complexity through a dedicated correction module
2Measurement precision
If extensive model training is performed to improve infilling accuracy, then measurement precision is improved, but loss of time increases
Solution Approach 1:
The system performs preliminary actions by pre-training generative models and preparing template-based approaches during an initial phase. Once these models are trained and the templates are established, they can be applied to new occlusion correction tasks without requiring extensive retraining, thus reducing the time loss for each individual correction operation while maintaining high infilling accuracy
Solution Approach 2:
The patent employs parameter changes by adjusting the level of model training and the selection of templates based on the specific characteristics of each occlusion case. This allows the system to optimize between training time and infilling accuracy by adapting parameters such as model complexity, template selection criteria, and inference settings to match the specific measurement and occlusion conditions
Data Source
AI summary
In variants, the method for occlusion correction can include: determining a measurement depicting an occluded object of interest (OOI), optionally infilling the occluded portion of the object of interest within the measurement, and determining an attribute of the object of interest based on the infilled measurement.