Video Facial Age Estimation With Matchability-Based Image Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video distribution systems face challenges in automatically detecting and prohibiting content based on the ages of individuals captured within videos, especially when the volume of content exceeds human moderation capabilities.
Innovation Solution
A system and method for age estimation that involves extracting frames from videos, performing facial detection, embedding facial images, clustering similar faces, and using a matchability algorithm to identify the best facial image for age classification, with optional third-party confirmation for accurate age estimation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If automated age estimation systems are implemented to detect underage content, then productivity in content moderation is improved, but measurement precision of age estimation may deteriorate due to reliance on algorithms rather than human verification
Solution Approach 1:
The system segments the age estimation process into multiple stages: initial automated age classification using machine learning models, followed by selective human review only for borderline or uncertain cases. This segmentation allows high-volume automated processing while preserving human precision for critical decisions.
Solution Approach 2:
The patent introduces an intermediary human review mechanism that acts as a mediator between automated age estimation and final content decisions. Human moderators review only the most uncertain cases, providing precision calibration without overwhelming computational resources.
2Measurement precision
If all facial images are processed for age estimation, then measurement precision is improved, but processing costs and computational resources increase
Solution Approach 1:
Instead of processing all facial images uniformly, the system applies partial action by selectively processing only those images that require it. The machine learning model initially screens all images, then only the most uncertain or borderline cases are sent for additional processing, reducing overall computational load while maintaining precision where needed.
Solution Approach 2:
The system applies different processing quality levels to different images based on their characteristics. High-quality detailed analysis is applied only to images with ambiguous age indicators, while standard processing is applied to clear cases, optimizing resource allocation based on local image characteristics.
Data Source
AI summary
Systems, methods, and computer-readable storage media for age estimation/classification, and more specifically to estimating/classifying the ages of people appearing within videos. Systems configured as disclosed herein can receive a video, then identify multiple facial images for each individual captured within the video. The system can create embeddings of the facial images, then cluster those images together based on distances between the corresponding embeddings. The system can also execute a matchability algorithm on those facial images, determining which of the images provides the clearest image of the individual(s), and can then estimate the age of the individual(s) using the best matchability images and/or send the best matchability image for each individual to a third party for analysis.