Human action recognition is used in many applications such as video\nsurveillance, human computer interaction, assistive living, and gaming. Many\npapers have appeared in the literature showing that the fusion of vision and\ninertial sensing improves recognition accuracies compared to the situations\nwhen each sensing modality is used individually. This paper provides a survey\nof the papers in which vision and inertial sensing are used simultaneously\nwithin a fusion framework in order to perform human action recognition. The\nsurveyed papers are categorized in terms of fusion approaches, features,\nclassifiers, as well as multimodality datasets considered. Challenges as well\nas possible future directions are also stated for deploying the fusion of these\ntwo sensing modalities under realistic conditions.\n