Detection performance of three models: Model A, Model B, Model C