Drowsiness Detection via Facial Landmark Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current drowsiness detection methods are either intrusive, affected by road conditions and driving speed, or have low accuracy, leading to ineffective alert systems for preventing accidents caused by drowsy driving.
Innovation Solution
A drowsiness detection system utilizing a combination of image sensors, movement sensors, and machine learning algorithms, including convolutional neural networks (CNNs), to track facial landmarks and vehicle movement, calculating PERCLOS, fixed gaze, and erratic movement to generate alarms when drowsiness is detected.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If physiological measurement devices are attached to the operator for drowsiness detection, then measurement accuracy is improved, but operator comfort and safety deteriorate due to intrusion and movement restriction
Solution Approach 1:
The patent replaces intrusive physiological measurement devices with a computer vision-based system using imaging devices (cameras) to capture facial images. Machine learning algorithms analyze facial features such as eye closure, yawning, and head position to detect drowsiness, eliminating the need for physical sensors attached to the operator's body.
Solution Approach 2:
The system introduces an intermediary computational layer (machine learning algorithms processing facial image data) between the operator and the detection system. This intermediary approach allows indirect measurement of physiological states through facial expressions and movements captured by imaging devices, avoiding direct physical contact with the operator.
2Difficulty of detecting and measuring
If steering wheel and vehicle movement sensors are used for drowsiness detection, then detection capability is improved, but reliability deteriorates due to interference from road conditions and driving speed
Solution Approach 1:
The patent extracts the detection focus from vehicle dynamics (steering wheel movement, vehicle acceleration) and shifts it entirely to the operator's facial features. By taking out the vehicle movement component and replacing it with facial image analysis, the system eliminates the confounding effects of road conditions and driving speed on detection accuracy.
Solution Approach 2:
Instead of inferring drowsiness from vehicle movement patterns (external indicators), the system inverts the approach by directly observing facial features (internal indicators of alertness). This inversion allows direct measurement of the operator's physiological state independent of vehicle dynamics or road conditions.
3Adaptability or versatility
If vehicle movement sensors are deployed for drowsiness detection, then detection coverage is improved, but response time deteriorates due to extended analysis period requirements
Solution Approach 1:
The system implements continuous periodic capture of facial images at high frame rates, enabling real-time analysis of drowsiness indicators. By using periodic imaging rather than continuous video processing, the system achieves rapid detection with reduced computational overhead, improving response time while maintaining comprehensive monitoring coverage.
4Measurement precision
If sensors are placed inside the steering wheel or dashboard for vehicle movement monitoring, then measurement accuracy is improved, but operator safety and comfort worsen due to potential interference with the driver
Solution Approach 1:
The system uses imaging devices to create visual copies (images) of the operator's facial features rather than using physical sensors that would require direct contact with the driver or vehicle controls. This copying approach allows accurate measurement of facial movements and expressions without any physical interference with the driver's operation of the vehicle.
Data Source
AI summary
A machine-implemented method for automated detection of drowsiness, which includes receiving from an imaging device directed at the face of an operator a series of images of the face of the operator onto processing hardware, on the processor detecting facial landmarks of the operator from the series of images to determine the level of talking by the operator, the level of yawning of the operator, the PERCLOS of the operator, on the processor detecting the facial pose of the operator from the series of images to determine the level of gaze fixation by the operator, on the processor calculating the level of drowsiness of the operator by ensembling the level of talking by the operator, the level of yawning of the operator, the PERCLOS of the operator and the level of gaze fixation by the operator, and generating an alarm when the calculated level of drowsiness of the operator exceeds a predefined value.