Face Tracking with Spectacles Proximity Constraints
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current face detection and tracking methods in video streams are less accurate and robust when an individual is wearing spectacles, as the lenses can deform or mask characteristic points, leading to incorrect face positioning and orientation.
Innovation Solution
A method that evaluates parameters of a face representation model, including proximity constraints between face and spectacles models, to improve pose estimation and tracking accuracy, using three-dimensional models and configuration parameters to account for spectacles' deformation and movement.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional face detection methods are used on faces wearing spectacles, then the detection process is simple and fast, but the detection precision and reliability deteriorate due to lens deformation and masking of characteristic points
Solution Approach 1:
The detection system is segmented into two independent modules: a spectacles detection module that detects spectacle frames and lenses, and a face detection module that detects facial characteristic points. This segmentation allows each module to specialize in detecting specific features, improving overall detection precision while managing complexity through modular design.
Solution Approach 2:
The detected spectacle model serves as an intermediary element that provides contextual information about the spectacles worn by the individual. This intermediary model is used to compensate for the masking and deformation effects of spectacles on face characteristic points, thereby improving face detection precision without requiring complete redesign of the detection system.
2Reliability
If face characteristic points are used for tracking, then the tracking method is computationally efficient, but the tracking reliability deteriorates when spectacles mask or deform these points
Solution Approach 1:
The tracking system merges two types of information: face characteristic point tracking and spectacle characteristic point tracking. By combining these two tracking streams, the system achieves higher reliability since the spectacle points provide additional stable reference features that are not affected by facial expressions or movements, while the computational load remains manageable through efficient point selection and matching algorithms.
3Measurement precision
If the detection system accounts for spectacles deformation and movement, then the tracking accuracy improves, but the computational complexity and processing time increase
Solution Approach 1:
The system performs preliminary detection of spectacle frames and lenses in advance, creating a spectacle model that accounts for deformation and movement characteristics. This preliminary action allows the main face tracking process to use pre-computed compensation information, improving pose estimation accuracy without requiring extensive real-time computations during the actual tracking phase.
Solution Approach 2:
The spectacle model is designed to be dynamic, automatically adapting to spectacles deformation and movement. The model parameters are updated based on detected spectacle positions and orientations, allowing the system to maintain high tracking accuracy throughout the video sequence without requiring re-detection of spectacle features, thus reducing processing time in subsequent frames.
Data Source
AI summary
A method is provided for tracking a face of an individual in a video stream acquired by an image-acquisition device, the face wearing a pair of spectacles. The method includes evaluating parameters of a representation of the face including a model of the pair of spectacles and a model of the face so that the representation of the face is superimposed on the image of the face in the video stream, the parameters being evaluated with respect to a plurality of characteristic points of the representation of the face, previously detected in an image of the video stream, referred to as first image, wherein all or some of the parameters of the representation are evaluated by taking account of at least one proximity constraint between at least one point of the model of the face and at least one point of the model of the pair of spectacles.