Video-based driver action recognition via hybrid spatial–temporal deep learning framework