Mask Head Pose Validation Study Dataset

Project Description Driver head pose is an important factor in transportation safety research. As the scale of naturalistic driving studies has grown, it has become increasingly important to develop automated methods for estimating driver head pose. The objective of this study was to evaluate the capability of the currently deployed Virginia Tech Transportation Institute (VTTI) head pose estimation system (mask). To this end the following tasks were completed: Identified conditions in which the mask system estimated head pose accurately and precisely. Determined if the mask system quality confidence measure accurately identified cases in which head pose estimates should not be used. Evaluated the ability of the mask system to predict eye glance location and provide a course estimate of head pose. Identified system limitations that should be considered by researchers using mask head pose data. Data Request Scope Information and some public data about the study can also be accessed on the InSight website Overview documentation can be accessed here . Data Specification The following data elements are available: Eye glance data Front, Rear, Hands and Face video streams Facial landmark annotations Geometric method head rotation estimates Mask head rotation estimates Vehicle kinematics