Pose-aware Multi-position Feature Network for Driver Distraction Recognition

Abstract This paper introduce pose-aware multi-position feature network for driver distraction recognition, taking into account the high association between visual and geometric features of neighboring human key points. method first uses object detector to detect driver’s body, then pose estimation points body. Finally, body are deconstructed obtain spatial features.In order create representation that corresponds each point, all chosen representations sent a convolutional neural network. The locations on upper used build reasoning module, which is transmitted get features. appearance fused with predict corresponding action through linear layers.The experimental result shows our methods achieve comparative performance own HY Large Vehicle Driver Dataset public AUC Distracted Dataset.