Efficient data enhancement optimization method in communication-assisted perception scene and related equipment

The invention provides an efficient data enhancement optimization method in a communication-assisted perception scene and related equipment. The method comprises the following steps: a data enhancement model receives perception data from a local end and performs data enhancement to obtain an enhanced sample; the deep reinforcement learning model receives local model parameters, constructs a current state by using each local model parameter and each reinforcement sample in each round, determines a current action in the current state, and determines a corresponding reward and a next state in the round; the current state, the current action, the corresponding reward and the next state form a transfer sample corresponding to the current state in the round, the transfer sample is utilized to train a deep reinforcement learning model, and the action comprises adding an enhanced sample to a corresponding local end; and the global model obtains local model parameters of each local end and aggregates the local model p