Neural Network Face Tracking via 2D Landmark Image Transformation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current face tracking techniques using two-dimensional facial landmark data are not compatible with neural networks, limiting their ability to predict face shape and expression activation effectively.
Innovation Solution
Transforming two-dimensional facial landmark data into image-based representations and inputting these representations into neural networks for training, allowing the networks to utilize the data for face tracking and expression recognition.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If two-dimensional facial landmark data is used directly as input, then the data structure is simple and easy to process, but the neural network cannot effectively predict face shape and expression activation
Solution Approach 1:
The patent introduces an intermediary representation (image-based landmark data) that bridges the gap between traditional 2D facial landmark data and neural network requirements. This intermediary format enables the neural network to effectively process and predict facial expressions while maintaining the essential information from the original landmark data.
Solution Approach 2:
The patent transforms the parameter representation of facial landmark data from a traditional 2D coordinate-based format to an image-based format. This parameter change allows the data to be compatible with neural network processing while preserving the essential facial feature information needed for accurate prediction.
2Adaptability or versatility
If traditional two-dimensional facial landmark data is used, then the processing approach is straightforward, but the compatibility with neural networks is poor
Solution Approach 1:
The patent introduces an intermediary representation (image-based landmark data) that bridges the gap between traditional 2D facial landmark data and neural network requirements. This intermediary format enables the neural network to effectively process and predict facial expressions while maintaining the essential information from the original landmark data.
Solution Approach 2:
The patent transforms the data from a 2D coordinate-based representation to an image-based representation, effectively adding a dimensional aspect that makes the data compatible with neural network processing architectures while preserving the essential facial feature information.
3Reliability
If image-based representation is created from landmark data, then neural network compatibility improves, but the processing time and computational resources increase
Solution Approach 1:
The patent performs the transformation from 2D landmark data to image-based representation as a preliminary step before neural network processing. This upfront transformation ensures that the data is in the correct format for accurate face tracking and expression prediction, avoiding the need for repeated transformations during processing.
Data Source
AI summary
Techniques for face tracking comprise receiving landmark data associated with a plurality of images indicative of at least one facial part. Representative images corresponding to the plurality of images may be generated based on the landmark data. Each representative image may depict a plurality of segments, and each segment may correspond to a region of the at least one facial part. The plurality of images and corresponding representative images may be input into a neural network to train the neural network to predict a feature associated with a subsequently received image comprising a face. An animation associated with a facial expression may be controlled based on output from the trained neural network.