In this paper, we study a structured signal detection problem in Ferromagnetic Ising models with examples encompassing Ising Models on lattices, and Mean-Field type Ising Models such as dense Erdős-Rényi, and dense random regular graphs. We provide sharp constants of detection in each of these cases and thereby pinpoint an asymptotically precise relationship between the detection problem with the underlying dependence. To obtain this sharp characterization of the detection boundary at the level of sharp multiplicative constants, we derive necessary moderate deviation bounds for partial summands of magnetizations which might be of independent interest. Finally, we demonstrate how our tests can be designed to be adaptive over the strength of dependence present in the respective models.