Microscope Scene Recognition Stabilization

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing microscope systems face challenges in accurately automating device settings for different scenes due to the need for extensive learning data and time, particularly with advanced functions, leading to potential hunting in inference results and frequent setting changes that hinder observation.

Innovation Solution

A microscope system that includes a scene recognition unit using a machine learning model to identify scenes, a determination unit to stabilize the recognition results over time, and a setting unit to adjust imaging settings based on the determined scene, thereby ensuring stable and appropriate device settings.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Extent of automation

If a learning model is used to automatically determine device settings based on live images, then automation of device settings is improved, but inference accuracy deteriorates leading to hunting and frequent setting changes

Engineering Contradiction:
Improveautomation of device settingsVSAvoidinference accuracy
Core Design Contradiction:
Extent of automationVSReliability

Solution Approach 1:

The system performs preliminary scene recognition on a captured image before using it for setting determination. By pre-processing the image through scene recognition, the system prepares more reliable input data for the learning model, thereby improving inference accuracy and reducing hunting phenomena while maintaining automation.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent introduces an intermediate processing step where captured images are first subjected to scene recognition to generate scene information, which then serves as input for the learning model. This intermediary process refines the input data quality, enabling more accurate and stable setting determinations without reducing automation.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Adaptability or versatility

If various device settings are supported in the microscope system, then adaptability to different scenes is improved, but the amount of learning data and time required increases

Engineering Contradiction:
Improvedevice settings for different scenesVSAvoidlearning time
Core Design Contradiction:
Adaptability or versatilityVSLoss of time

Solution Approach 1:

The system segments the scene recognition process into distinct stages: capturing an image, performing scene recognition to identify scene type, and then determining device settings based on the recognized scene. This segmentation allows the use of pre-trained scene recognition models for common scene types, reducing the need for extensive custom learning data while maintaining adaptability to various microscopy scenarios.

Inventive Principle:
Principle #1Segmentation

Data Source

PatentEP4530695A1Microscope system, setting changing method, program, and digital camera
Publication Date: 2025.04.02 EVIDENT CORP
  • EP4530695A1 patent drawingFigure 1~2
  • EP4530695A1 patent drawingFigure 3~4
  • EP4530695A1 patent drawingFigure 5

AI summary

A microscope system includes: a microscope; a digital camera configured to image a sample through the microscope; a scene recognition unit configured to perform scene recognition based on an image of the sample obtained by the digital camera, using a machine learning model that has learned a plurality of scenes; a scene determination unit configured to perform scene determination based on a recognition result of the scene recognition unit; a stabilization unit configured to temporally stabilize a determination result of the scene determination unit; and a setting unit configured to change settings of the digital camera based on the determination result of the scene determination unit.