Driver Gaze Classification for Low-False-Alarm Distraction Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing driver monitoring systems (DMS) face challenges in accurately detecting driver distraction and reducing false alarms, particularly due to varying installation locations and driver behaviors, which can lead to decreased accuracy and increased false alarms.
Innovation Solution
The proposed method involves using a DMS system that receives images of the driver and employs a pre-trained model to determine the driver's gaze class by extracting yaw and pitch values. The system then calculates a distraction score based on the distance between these values and predefined gaze clusters, and generates an alarm only when the score exceeds a certain threshold.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a sensitive driver detection algorithm is used to improve detection accuracy, then the accuracy of distraction detection is improved, but the frequency of false alarms increases
Solution Approach 1:
The detection process is divided into multiple independent stages: initial distraction detection, separate alarm condition detection, and integrated alarm generation. By segmenting the detection logic, the system can evaluate multiple conditions before triggering an alarm, reducing false positives while maintaining high detection sensitivity
Solution Approach 2:
The system incorporates feedback mechanisms where detection results from multiple algorithms are combined and evaluated together. The integrated alarm generation uses feedback from both the distraction detection algorithm and the separate alarm algorithm to make the final determination, allowing the system to learn from and adjust to various driving scenarios
2Adaptability or versatility
If DMS products are customized for each vehicle to adapt to different installation locations, then the adaptability to different vehicles is improved, but the device complexity and customization requirements increase
Solution Approach 1:
The detection system is designed with universal algorithms that can function across different vehicle types and installation locations. The gaze cluster generation and distraction detection methods are vehicle-agnostic, allowing a single system design to serve multiple applications without requiring extensive customization for each vehicle
Solution Approach 2:
The system adapts to different vehicles by adjusting parameters such as gaze clusters and detection thresholds rather than redesigning the entire system. By changing key parameters like the gaze cluster database and alarm thresholds, the same core algorithm can be optimized for different vehicle configurations and installation positions
Data Source
AI summary
A method for detecting distraction of a driver in a driver monitoring system is disclosed. The method includes receiving a first image including the driver within a vehicle. The method includes determining a gaze class corresponding to the first image among a plurality of gaze classes by using a first model in response to the receiving the first image. The plurality of gaze classes includes a first gaze class indicating that the driver is gazing in a front direction and a second gaze class indicating that the driver is gazing in a non-front direction. The method includes determining whether the distraction of the driver is present in the first image based on the gaze class corresponding to the first image.