Circadian Rhythm Calibration for Vision-Based Drowsiness Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing driver drowsiness detection systems in vehicles suffer from inaccuracies due to factors like occluded views of the driver, difficult imaging conditions, and biased self-reported ground truth data, leading to unreliable drowsiness estimates.
Innovation Solution
Incorporating circadian rhythm data to augment vision-based assessments by estimating the driver's position on the circadian cycle, using factors like time of day, sleep quality, and recent sleep patterns to correct and calibrate drowsiness scores, thereby generating more accurate ground truth training data for machine learning models and improving real-time drowsiness detection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If vision-based techniques are used to detect driver drowsiness, then real-time monitoring capability is improved, but measurement precision deteriorates due to occluded views and difficult imaging conditions
Solution Approach 1:
The patent introduces circadian rhythm data as an intermediary element that mediates between vision-based detection and ground truth drowsiness levels. This additional data source helps correct inaccuracies from occluded views by providing physiological context about expected drowsiness levels at different times of day, thereby improving measurement precision while maintaining real-time monitoring capability
Solution Approach 2:
The system combines multiple data sources (vision-based metrics, circadian rhythm data, and ground truth labels) into a composite training dataset. This composite approach integrates information from different modalities to compensate for the weaknesses of individual sources, particularly addressing measurement precision issues caused by occluded views through the addition of circadian rhythm context
2Ease of manufacture
If self-reported ground truth data is used for training machine learning models, then data collection ease is improved, but reliability deteriorates due to biased reporting
Solution Approach 1:
The patent implements a feedback mechanism where circadian rhythm data is used to verify and correct self-reported ground truth labels. The system compares expected drowsiness levels (based on circadian phase) with self-reported levels, identifying and correcting biased reports while maintaining the ease of data collection through self-reporting
Solution Approach 2:
The system uses automatically collected circadian rhythm data to self-correct the training dataset without requiring additional manual verification of each self-reported label. This self-service approach maintains data collection ease while improving reliability through automated bias detection and correction using physiological data
3Measurement precision
If circadian rhythm data is incorporated to correct drowsiness assessments, then measurement precision is improved, but device complexity increases
Solution Approach 1:
The patent applies circadian rhythm data serving multiple functions: it corrects ground truth labels during training data preparation, validates machine learning model predictions during deployment, and provides contextual information for improving measurement precision. This multi-functionality justifies the added complexity by delivering comprehensive improvements across different system stages
Solution Approach 2:
The system modifies the parameter space by adding circadian phase information to the existing drowsiness detection framework. This parameter change enables more accurate assessments without fundamentally redesigning the entire system, as the circadian data works alongside existing vision-based metrics rather than replacing them
4Productivity
If machine learning models are trained with corrected ground truth data, then productivity is improved through more accurate detection, but loss of information increases due to discarding potentially inaccurate self-reported labels
Solution Approach 1:
The patent replaces the mechanical process of manually verifying each self-reported ground truth label with an automated computational process that uses circadian rhythm data to identify and correct biased reports. This substitution maintains productivity by automating the correction process while reducing information loss through systematic rather than arbitrary data filtering
Data Source
AI summary
In various examples, circadian rhythm-based data augmentation for drowsiness detection systems and applications are provided. Embodiments described herein may produce an estimated circadian rhythm for a test subject and/or vehicle driver or other machine operator or occupant, and use the pattern of that circadian rhythm to correct, confirm, calibrate, or otherwise augment drowsiness assessments derived from video image data. The position of a person in the context of their process C circadian cycle may be used as indication of their level of drowsiness. An estimated process C circadian cycle may be used to generate more accurate ground truth training data for training machine learning models, and may be used by real-time, in-vehicle drowsiness detection systems that infer driver drowsiness levels based on captured images. In various embodiments, a circadian rhythm drowsiness estimate may be used to correct, calibrate, augment, and/or replace a drowsiness score predicted by a machine learning model.