A model of indoor scene video recognition technology applied to home robots is proposed based on visual attention model. This model creates image features based on the feature map of visual attention model by combining with spatial information. In classification, complex details of the indoor image are avoided and spatial information of the different scenes are retained by applying these features. Moreover, a home service robot is used as experimental platform to conduct experiments. The result shows that good recognition results can be achieved by using the proposed model, and real-time video recognition are implemented on the robot.