Image Processing Device for Surgical Data Prioritization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing systems in surgical environments face challenges in efficiently obtaining data that contributes to improved recognition performance due to the inability to determine the importance of images in real-time, leading to inefficient data collection for enhancing recognition performance.
Innovation Solution
An image processing device and method that extracts intermediate feature amounts from input images, calculates their importance, and stores them based on this importance, allowing for the efficient collection of data that enhances recognition performance by prioritizing images with high importance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If additional labeling is performed to improve recognition performance, then recognition performance improves, but time consumption and operational complexity increase
Solution Approach 1:
The system automatically determines image importance and performs selective accumulation without requiring manual labeling or intervention. The recognition model itself evaluates its own learning needs by analyzing intermediate feature amounts, enabling the system to self-identify which images should be accumulated for future learning, thus eliminating the need for external labeling efforts.
Solution Approach 2:
The system uses the output from the recognition model's intermediate feature extraction as feedback to determine image importance. This feedback loop allows the system to continuously identify which images would most benefit future learning iterations, enabling automatic prioritization of data accumulation without manual intervention while improving recognition performance over time.
2Reliability
If all images are accumulated for learning, then recognition performance may improve, but storage requirements and processing complexity increase
Solution Approach 1:
Instead of treating all images uniformly, the system applies different quality standards to different images based on their importance scores. Images are selectively accumulated based on their local importance characteristics determined by the recognition model's intermediate feature analysis, allowing the system to focus resources on high-value images rather than treating all images equally.
Solution Approach 2:
The system changes the parameter of image selection from binary (accumulate or discard) to a continuous importance score based on intermediate feature amounts. This parameter transformation enables nuanced decision-making about which images to accumulate, optimizing the balance between recognition performance improvement and storage/resource management complexity.
3Productivity
If importance calculation is performed in real-time, then efficient data collection is achieved, but computational load increases
Solution Approach 1:
The system performs importance calculation only on the necessary intermediate feature amounts from the recognition model, rather than analyzing all possible image characteristics. This partial action approach focuses computational resources on the specific features that matter for determining learning value, achieving real-time importance assessment without exhaustive analysis.
Solution Approach 2:
The importance calculation leverages intermediate feature amounts that are already extracted during the normal recognition process. By reusing these pre-computed features for importance determination, the system avoids redundant calculations and performs importance assessment with minimal additional computational overhead, enabling real-time operation.
Data Source
AI summary
An image processing device (40) of a form according to the present disclosure includes: a feature amount extraction section (42a) that extracts an intermediate feature amount related to machine learning from an input image that is an image inside a body; an importance calculation section (42b) that calculates image importance of the input image on the basis of the intermediate feature amount; and an image accumulation section (42c) that stores the input image on the basis of the image importance.