Vision Assistance Device Using Learning Model for Visual Impairment Adaptation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current vision assistance technologies for visually impaired individuals and those with visual impairments due to diseases like glaucoma struggle to provide accurate information about the scene in front of them, as they rely solely on sensor data analysis without effectively accounting for different types of visual impairments and their varying effects.
Innovation Solution
A vision assistance apparatus that captures images using an image acquisition unit, analyzes them with a learning model to extract features, and provides notification signals through a control unit, offering customizable assistance modes such as total blindness, low vision, glaucoma, macular hole, strabismus, and auditory assistance, adjusting output accordingly to suit individual needs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If vision assistance technologies rely solely on sensor data analysis, then device complexity is reduced, but measurement precision and information accuracy deteriorate
Solution Approach 1:
The patent merges sensor data analysis with image capture and analysis technologies. The vision assistance device integrates multiple data sources including sensors and image capture units to analyze the scene in front of the user, thereby improving measurement precision while managing device complexity through unified processing in the controller.
Solution Approach 2:
The controller performs multiple functions: it processes sensor data, analyzes captured images using learning models, generates notification signals, and provides vision assistance. This multi-functionality allows the device to achieve high measurement precision through comprehensive analysis while avoiding excessive complexity by centralizing processing in a single controller.
2Device complexity
If a single vision assistance technology is used for all visual impairments, then device complexity is reduced, but adaptability deteriorates
Solution Approach 1:
The patent implements dynamic adaptability through multiple operation modes (total blindness mode, low vision mode, glaucoma mode, macular hole mode, strabismus mode) that can be selected based on the user's specific visual impairment type. The controller dynamically adjusts its processing and notification strategies according to the selected mode, enabling the device to adapt to different conditions without requiring multiple separate devices.
Solution Approach 2:
The vision assistance device changes operational parameters based on the selected operation mode. For example, in total blindness mode, the device provides audio notifications, while in low vision mode, it processes and displays images with adjusted parameters. This parameter adjustment allows the single device to serve multiple impairment types effectively.
3Speed
If image analysis is performed without learning models, then processing speed is improved, but measurement precision deteriorates
Solution Approach 1:
The patent applies preliminary action by pre-training learning models offline before actual use. The controller executes these pre-trained learning models during operation to quickly classify objects and extract features from captured images. This approach enables fast processing during actual use while maintaining high measurement precision, as the complex learning computations were performed in advance.
Data Source
AI summary
A vision assistance apparatus may include an image acquisition unit configured to acquire an image by capturing the scene of the front which a user watches, a sensor unit configured to acquire sensing information on objects located in front of the user, a control unit configured to analyze the image acquired by the image acquisition unit and generate a notification signal for the front scene through an analysis result of the image and the sensing information acquired by the sensor unit, and an output unit configured to provide the user with the notification signal generated by the control unit in the form of sound.