Image Capturing Apparatus Learning User Preferences
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional life-log cameras struggle to capture desired moments automatically due to the infinite variety of objects and environments, making it difficult to compensate for shortages in learning patterns using image processing, and often record unwanted videos.
Innovation Solution
An image capturing apparatus equipped with an acquisition circuit, a learning circuit, and a control circuit that learns user preferences based on supervised data to automatically adjust image capturing settings, including data registration from user-instructed captures and pre/post-capture images, to optimize video acquisition.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If automatic image capturing is performed at predetermined time intervals, then image capturing can be automated without user operation, but unwanted videos are recorded and desired moments are missed
Solution Approach 1:
The system uses user feedback from manual capturing operations to learn and improve automatic capturing decisions. The learning circuit analyzes supervised data including images captured by user instruction and pre/post images to understand user preferences, enabling the control circuit to make more accurate automatic capturing decisions that align with user expectations.
Solution Approach 2:
The system captures pre-images and post-images around manual capturing moments to proactively gather learning data. By capturing images before and after user-instructed captures, the system prepares supervised data in advance that helps the learning circuit understand the context and characteristics of desired moments, improving future automatic capturing accuracy.
2Measurement precision
If learning function is added to capture desired moments, then automatic capturing accuracy improves, but enormous amount of supervised data is necessary
Solution Approach 1:
The system proactively captures pre-images and post-images around manual capturing operations to accumulate learning data in advance. This preliminary action ensures that supervised data is gathered systematically over time, providing the learning circuit with sufficient training material to improve automatic capturing accuracy without requiring external data sources.
Solution Approach 2:
The same image capturing device serves multiple functions: it captures images for user viewing, captures pre/post images for learning, and performs automatic capturing. This multi-functionality allows the system to accumulate supervised data using the existing hardware without requiring additional data collection mechanisms, efficiently utilizing the device's capabilities to generate training data.
3Productivity
If life-log camera is worn on body for continuous capturing, then user's daily life scenes are recorded, but unwanted videos are acquired and storage space is wasted
Solution Approach 1:
The system uses feedback from user manual capturing operations to learn what moments are valuable to the user. By analyzing patterns in user behavior and the characteristics of captured images, the control circuit can identify and capture only the moments the user actually wants to preserve, eliminating unnecessary recordings and optimizing storage space efficiency.
Solution Approach 2:
Instead of continuously capturing all moments, the system performs selective capturing based on learned user preferences. The control circuit decides when to capture images by evaluating current conditions against learned patterns, performing capturing only when necessary to record desired moments, thus reducing overall capturing frequency while maintaining recording quality.
Data Source
AI summary
An image capturing apparatus includes an acquisition unit configured to acquire data concerning a captured image captured by an image capturing unit, a learning unit configured to learn a condition of an image that a user likes, based on supervised data, a control circuit configured to decide automatic image capturing by the image capturing unit based on the condition learned by the learning unit and to register, as the supervised data, data acquired by the acquisition unit for a captured image obtained by image capturing performed based on an instruction of the user, and data acquired by the acquisition unit for captured images for learning which are captured before and/or after the image capturing performed based on an instruction of the user.