The present disclosure discloses a method for improving recognition accuracy of object attributes. The method is performed by a target node and includes: obtaining first attribute data of a target object in a sensing region; in response to the first attribute data failing to satisfy a predetermined sensing condition, querying nodes by the target node to determine N target auxiliary nodes matching the first attribute data, where N is a positive integer greater than or equal to 1; and, obtaining data from the N target auxiliary nodes, where the data includes second attribute data of the target object in the sensing region, and an accuracy of the second attribute data is higher than an accuracy of the first attribute data.