Target Identification via Impurity-Driven Question Generation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional techniques for identifying a final target among multiple persons using image features are inefficient as they do not effectively utilize information beyond user answers, particularly when presuming a user from captured images.
Innovation Solution
An information processing apparatus and mobile object equipped with an image acquisition unit, extraction unit, impurity acquisition unit, and generation unit that detect and extract features from images, calculate impurity to determine the inseparability of targets, and generate questions to minimize the number of questions needed to accurately identify the final target.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If conventional question generation techniques are used to identify a final target among multiple persons, then the system can narrow down candidates through user answers, but the number of questions required is large and the process is inefficient
Solution Approach 1:
The patent merges image recognition results with question generation by integrating feature extraction from captured images directly into the question generation process. The image recognition unit extracts features such as clothing color, hair color, and other visual characteristics, which are then combined with the question generation unit to create more targeted and efficient questions, reducing the total number of interactions needed for identification.
Solution Approach 2:
The system performs preliminary action by pre-extracting visual features from images before the questioning process begins. The image recognition unit captures and analyzes visual characteristics of multiple persons in advance, storing these features for use during question generation. This preliminary feature extraction allows the system to start with more informed questions, reducing the overall identification process time.
2Loss of information
If only user answers are used for target identification, then the process is simple, but information from captured images is not effectively utilized
Solution Approach 1:
The system implements multi-functionality by making the image recognition unit serve multiple purposes: it not only captures images but also extracts relevant visual features that feed into the question generation process. The same image processing capability is used both for initial target detection and for providing contextual information throughout the questioning process, maximizing the utility of the image data without requiring separate dedicated systems.
Solution Approach 2:
The patent introduces an intermediary mechanism in the form of the feature extraction process that bridges image recognition and question generation. Rather than directly using raw images or simple user answers, the system extracts intermediate visual features (such as clothing characteristics, physical attributes) that mediate between the image data and the questioning logic, enabling effective information integration while maintaining manageable system complexity.
3Reliability
If multiple questions are asked to ensure accurate identification, then the reliability of target presumption improves, but the user experience deteriorates due to excessive questioning
Solution Approach 1:
The system applies local quality by focusing question generation on specific, locally-relevant features extracted from images of individual persons. Rather than asking generic questions that apply to all candidates, the system identifies which visual features (such as a particular clothing item or physical characteristic) are most distinctive for each person and generates questions targeted at those specific local characteristics, improving accuracy while reducing the total number of questions needed.
Data Source
AI summary
An information processing apparatus of the present invention comprises acquires a captured image; detects a plurality of targets included in the captured image, and extracts a plurality of features for each of the detected plurality of targets; acquires an impurity for each extracted feature, the impurity indicating a degree to which a predetermined target is inseparable from among the plurality of targets in a case where a user is asked a question for presuming the predetermined target from among the plurality of targets based on each feature; and generates the question to reduce a number of questions for minimizing the impurity based on the extracted features and the impurity for each of the features.