Biometric Pattern Recognition TRIZ Cases
- Non-Rigid Skeleton Model for Human Body Representation
Decoupling bone length variations from the geometric skeleton reduces model complexity while maintaining visual realism in human body representations.
- Touchscreen Gesture Recognition via Segmented Touch Zones
A touchscreen divides the display into distinct touch zones to recognize simultaneous multi-hand inputs without accidental single-hand detection.
- Occupancy Count Correction via Beta-Binomial Regression
Beta-binomial regression models historical ingress and egress data to correct real-time exit counts, resolving sensor noise and hardware aging inaccuracies.
- Context-Based Object Clustering via Tagged Attribute Mapping
Tagging non-physical attributes to physical ones resolves low precision in keyword clustering by enabling dynamic, context-based grouping.
- Deep Learning Pig Weight Estimation via Keypoint Detection
Keypoint detection and ResNext networks estimate pig weight from 2D images, resolving nonlinear data processing limits in manual measurement.
- Augmented Reality Guest Recognition for Real-Time Experience Personalization
An augmented reality system identifies guests and displays personalized experience data to venue staff.
- Object Detection in Electronic Devices Using Optical Feature Evaluation
Evaluating optical features such as hair gloss and pupil brightness resolves face detection accuracy issues caused by planar or three-dimensional obstructions.