Facial Detection Using Assurance and Connection Factors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image management tools inaccurately identify objects with facial characteristics as human faces, leading to false hits and inefficiencies in facial recognition processes.
Innovation Solution
A multi-tiered facial detection method involving assurance factor and connection factor generators to evaluate and filter facial regions, ensuring accurate identification by assigning assurance factors based on facial characteristics and connection factors between regions, and clustering to discard non-human facial regions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If automatic facial detection is used in image management tools, then the process of identifying individuals in images is automated and simplified, but false hits occur where objects with facial characteristics are incorrectly identified as human faces
Solution Approach 1:
The patent segments the facial detection process into multiple independent stages: detection stage (identifying possible facial regions), verification stage (assigning assurance factors based on facial characteristics), and clustering stage (grouping and validating faces). This segmentation allows each stage to focus on specific aspects, improving overall reliability by filtering out false hits at intermediate stages.
Solution Approach 2:
The patent introduces intermediate verification mechanisms between detection and final identification. Assurance factors serve as intermediaries to evaluate the quality of detected facial regions, while connection factors act as intermediaries to assess relationships between detected faces. These intermediaries prevent direct acceptance of all detected regions, reducing false hits while maintaining automation.
2Reliability
If manual organization of photos is performed to avoid false hits, then accuracy of facial identification is improved, but the process becomes tedious and time-consuming
Solution Approach 1:
The system performs self-verification of detected faces through automated assurance factor assignment and clustering. Instead of requiring manual verification, the system uses computational methods to automatically evaluate facial characteristics, assign confidence scores, and group similar faces, thereby maintaining high accuracy while eliminating the time-consuming manual process.
Solution Approach 2:
The patent transforms the verification process from manual to automated by changing the evaluation parameters to computational metrics. Assurance factors and connection factors are calculated using image processing algorithms that analyze facial characteristics, replacing manual visual inspection with automated parameter-based evaluation that is both accurate and time-efficient.
3Productivity
If all possible facial regions are processed without filtering, then the detection process is simple and fast, but false hits increase and reliability decreases
Solution Approach 1:
The patent applies preliminary filtering actions during the detection process itself. Instead of processing all detected regions equally, the system performs preliminary verification by assigning assurance factors to filter out low-quality detections early. This preliminary action reduces the number of false hits before final processing, maintaining speed while improving reliability.
Solution Approach 2:
The patent applies different quality standards to different detected facial regions based on their local characteristics. Regions with higher assurance factors (indicating stronger facial characteristics) are processed with higher priority, while regions with lower assurance factors are filtered out or processed less extensively. This local quality differentiation maintains processing efficiency while improving overall accuracy.
Data Source
AI summary
Various embodiments are disclosed for detecting facial regions in a plurality of images. In one embodiment, a method comprises assigning at least one of the possible facial regions an assurance factor, forming clusters of possible facial regions based on a connection factor between the facial regions, and determining facial regions from the possible facial regions based on the assurance factor and the clusters of possible facial regions.