Helicopter wire strikes have been one of the most common sources of accidents in the past seven years. Poles and wires can be difficult to see for rotorcraft pilots because of their size, their location, and their blending into the background. This motivates the need to provide pilots with more tools and information about wire and pole locations. However, there is currently no comprehensive dataset summarizing the locations of poles and wires. As such, this work aims to leverage various types of data sources to generate a database of wires of interest. The proposed approach consists in first detecting pole and wire locations and then predicting a wire network. The first step combines a semantic segmentation approach applied to satellite imagery and a convolutional neural network classifier applied to street-view imagery to predict wire and pole locations. The second step applies the many-to-many Dijkstra algorithm to connect the predicted locations and generate a predicted wire map. This two-step approach is applied to a portion of Westchester County in New York, where publicly available imagery exists. Results indicate that this approach can successfully predict a wire map with high accuracy for wires alongside roads.