Context-Aware Notification Using Emotion and Activity Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems fail to timely notify users of necessary activities based on their current state and situation, particularly when unscheduled events occur, leading to unnecessary notifications or missed important tasks.
Innovation Solution
An information processing apparatus that analyzes captured images to estimate a user's emotion, activity, and situation, determining appropriate implementing items using learning models, and sends tailored notifications through a notification unit.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If implementing items are registered in association with time periods in advance, then the system can notify users of scheduled activities, but the system cannot identify or notify of unscheduled living activities
Solution Approach 1:
The system performs preliminary registration of implementing items with time periods, but also continuously monitors actual user behavior without time constraints. When unscheduled activities are detected through image capture and emotion recognition, the system dynamically adds notification items beyond the pre-registered schedule, enabling both scheduled and unscheduled activity monitoring.
Solution Approach 2:
The notification system transitions from a static pre-registered schedule to a dynamic structure that adapts to real-time user behavior. The estimation unit continuously analyzes captured images and emotion data to determine current user state, allowing the notification unit to dynamically select and notify of appropriate implementing items based on actual situation rather than fixed time periods.
2Ease of operation
If implementing items are determined in advance, then the system can provide structured notifications, but the system may notify users of items that should not be implemented at that time, causing annoyance
Solution Approach 1:
The system changes the parameters for selecting implementing items based on real-time emotion recognition results. When the estimation unit detects specific emotions or user states from captured images, the notification unit adjusts which implementing items are selected for notification, ensuring only relevant items are notified at appropriate times rather than following a fixed predetermined list.
Solution Approach 2:
The system incorporates feedback loops where the estimation unit continuously monitors user emotion and state from captured images, and the notification unit uses this feedback to dynamically adjust notifications. This feedback mechanism ensures notifications are tailored to current user conditions, preventing annoyance from irrelevant notifications while maintaining convenience for relevant ones.
3Quantity of substance
If the system notifies users of all pre-registered implementing items, then the system provides comprehensive coverage, but the system generates unnecessary notifications that annoy users
Solution Approach 1:
Instead of uniformly notifying about all implementing items, the system applies local quality by selectively notifying only about specific items based on real-time emotion recognition. The estimation unit analyzes captured images to determine current user state, and the notification unit then targets only the relevant implementing items for that specific situation, reducing unnecessary notifications while maintaining comprehensive coverage for actually needed items.
Data Source
AI summary
The technology of the present disclosure is capable of notifying of an implementing item which is considered necessary in a timely manner based on the target person's state and situation. A person recognition unit performs person recognition processing on the obtained captured image by using a learning model. An emotion recognition unit estimates the emotion of the person in the obtained captured image by performing emotion recognition processing on the image of the person. An activity prediction unit predicts the activity of the person by performing activity prediction processing on the obtained captured image. A situation recognition unit recognizes the target person and the situation around that person by using a learning model that has learned. An implementing item determination unit determines one or more implementing items to notify of based on the obtained estimation result and implementing item, implementation conditions, and notification destinations are associated with one another.