Abstract: Human-object interaction recognition is a challenging problem as it is a combination of three challenging tasks in computer vision, namely human-action recognition, object detection and the scene understating. These tasks share many challenges such as the appearance of a human performing a specific action can be a rich source of information and indication about the type of the performed action. other challenges such as occlusions, the layout of the scene, variation of body pose, and object appearance make it very important to understand to distinguish between two similar actions. The scope of this study is limited to actions were humans interacting with objects. Therefore, we introduce a new taxonomic classifications for human-object interaction. Also, we present a number of approaches that have been introduced recently that can be applied to a real-word applications. Finally, we present a number of human-object interaction datasets that are publicly available.