Image Tagging System Using Time-Interval Segmentation for Facial Change Tracking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current image automatic tagging methods struggle to track and identify individuals as they age, as features change over time, making it difficult to automatically classify and search for images of people across different periods, especially in vast image pools like the internet.
Innovation Solution
User equipment and methods that classify and tag images based on time intervals, updating feature information of reference images by calculating similarities with images from the same and adjacent time intervals, allowing for the tracking and identification of individuals despite facial changes over time.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If simple classification criteria are used for automatic image tagging, then the tagging process is fast and easy to implement, but the accuracy of tracking individuals across different time periods deteriorates
Solution Approach 1:
The patent segments the image database into multiple time intervals and performs tagging operations within each segment. By dividing the large-scale image set into manageable time-based segments, the system can efficiently process images while maintaining accurate individual tracking through localized feature comparison within each time interval.
Solution Approach 2:
The system performs preliminary tagging within each time interval before conducting cross-time interval matching. This preliminary action prepares feature information in advance, enabling faster subsequent matching operations while maintaining high accuracy through multi-stage processing.
2Measurement precision
If feature information is updated by comparing with all images in the database, then individual tracking accuracy improves, but the computational complexity and time required increase significantly
Solution Approach 1:
The patent divides the image database into multiple time intervals and limits feature comparison operations to within each time interval. This segmentation reduces the computational burden from comparing against all images to comparing only against relevant time-bound images, while still achieving accurate individual tracking through cumulative updates across intervals.
Solution Approach 2:
The system performs partial updates of feature information by only comparing with images from specific time intervals rather than all images. This partial action is sufficient for maintaining tracking accuracy while significantly reducing computational complexity compared to exhaustive comparison.
3Measurement precision
If the system searches for exact matches of reference images, then identification precision is high, but the system cannot adapt to facial changes over time
Solution Approach 1:
The patent implements dynamic feature information that is continuously updated across time intervals. Instead of using static reference image features, the system dynamically adjusts feature representations by incorporating information from multiple time periods, enabling adaptation to facial changes while maintaining identification precision through cumulative learning.
Solution Approach 2:
The system changes the parameters of feature information over time by updating them with data from successive time intervals. This parameter evolution allows the system to adapt to facial changes while maintaining high identification precision through progressive refinement of feature representations.
4Reliability
If the system processes all images simultaneously for tagging, then comprehensive coverage is achieved, but processing time and resource consumption increase
Solution Approach 1:
The patent segments the image processing task into multiple time interval-based batches. By processing images in segmented time intervals rather than all at once, the system achieves comprehensive tagging coverage while significantly reducing processing time and resource consumption through distributed, incremental computation.
Solution Approach 2:
The system performs preliminary tagging within each time interval before proceeding to the next interval. This preliminary action within segments enables comprehensive coverage to be achieved incrementally, reducing overall processing time while maintaining complete coverage through systematic progression through all image sets.
Data Source
AI summary
Provided are user equipment, a control method thereof and a non-transitory computer readable storage medium having a computer program recorded thereon. That is, according to the present invention, an image at a specific time is searched by identifying and tracking a person included in a corresponding reference image among a plurality of images photographed at a predetermined time different than a photographing time of the corresponding reference image by using the reference image, and a search result is provided or the images are classified according to the search result to conveniently acquire a desired image search result through the reference image, thereby improving convenience of a user.