首页|Patent Application Titled 'Personalized Calibration Functions For User Gaze Dete ction In Autonomous Driving Applications' Published Online (USPTO 20240143072)
Patent Application Titled 'Personalized Calibration Functions For User Gaze Dete ction In Autonomous Driving Applications' Published Online (USPTO 20240143072)
扫码查看
点击上方二维码区域，可以放大扫码查看
原文链接
NETL
NSTL
Reporters obtained the following quote from the background information supplied by the inventors: “Autonomous and semi-autonomous vehicles leverage machine lear ning approaches-such as those using deep neural networks (DNNs)-to analyze image s of a user to determine various information about a user’s gaze. This gaze info rmation may then be used to take responsive action to prevent harm to the user, such as redirecting the user’s attention to a potential hazard, pulling the vehi cle over, and/or the like. For example, DNNs may be used to detect that a driver is falling asleep at the wheel, based on the user’s downward gaze toward the fl oor of the vehicle, which may lead to an adjustment in the speed and direction o f the car (e.g., pulling the vehicle over to the side of the road) or an auditor y alert to the driver. However, determining a gaze of a person can be difficult, given that each individual has such different measurements with respect to thei r eyes (e.g., angular offset between visual and optical axes of an eyeball, corn ea radius and curvature, distance between pupil center and corneal center, refra ction of aqueous humor and cornea) and/or other facial features in addition to d ifferent heights, seating positions, and/or other posture information that may a ffect the accuracy of the DNNs deployed to detect the user’s gaze.
NETL
NSTL
