Facial Region Scoring for Subject-Crowd Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current facial recognition systems in photography fail to distinguish between subject faces and crowd faces, leading to incorrect categorization and requiring manual user intervention to remove unwanted faces, resulting in a negative user experience.
Innovation Solution
The system employs facial feature analysis using algorithms like modified ORB, combined with k-means clustering, to score and filter out crowd faces from subject faces, automatically categorizing images based on user intent and context.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If facial recognition systems detect all faces in images, then the quantity of identified persons increases, but the accuracy of subject identification decreases due to inclusion of crowd faces
Solution Approach 1:
The patent segments faces into two categories: subject faces and crowd faces. This is achieved by analyzing facial features and determining a score for each facial region, then using clustering algorithms to separate faces into distinct groups. The segmentation allows the system to identify multiple faces while maintaining accuracy by distinguishing which faces are actual subjects versus background crowd members.
Solution Approach 2:
The patent applies different analysis criteria to different faces within the same image based on their local characteristics. By evaluating facial features individually and assigning scores to each facial region, the system treats each face locally rather than applying a uniform threshold. This local quality approach enables accurate differentiation between subject and crowd faces even when multiple faces are present.
2Productivity
If existing facial recognition systems categorize all detected faces, then image categorization coverage improves, but user experience deteriorates due to manual removal of unwanted crowd faces
Solution Approach 1:
The patent implements self-service by enabling the system to automatically distinguish and categorize subject faces versus crowd faces without user intervention. The facial feature analysis and clustering algorithms autonomously identify which faces should be included in image categorization, eliminating the need for users to manually review and remove unwanted crowd faces while maintaining comprehensive categorization coverage.
3Quantity of substance
If crowd faces are included in image sets, then the completeness of people categorization improves, but the reliability of facial recognition decreases due to incorrect identification
Solution Approach 1:
The patent performs preliminary action by analyzing facial features and determining scores for each facial region before final categorization. The clustering algorithm pre-separates faces into subject and crowd groups, allowing the system to maintain complete categorization of all detected faces while ensuring high reliability by pre-identifying which faces are genuine subjects versus crowd members before generating the final image sets.
Data Source
AI summary
Systems and methods disclosed herein for providing people categorization within electronic images. One embodiment involves retrieving an input image from memory. The embodiment further involves determining one or more facial regions in the input image using a facial detection algorithm. The embodiment further involves identifying a number of features for each facial region. The embodiment further involves determining a subject face and a crowd face in the input image based on at least the number of features for the subject face being more than the number of features for the crowd face. The embodiment further involves displaying, on a display device, the subject face identified within the image.