Electricity powers our daily lives, but behind this constant availability lies an extensive infrastructure. Power generated at plants is transmitted over long distances via conductor wires supported by towers of various shapes and designs. One common type is the H-frame structure, which can suffer damage over time, with each component exhibiting distinct modes of degradation. Detecting these issues early helps prioritize maintenance, preventing energy loss, further infrastructure damage, and supply disruptions.

Power companies routinely inspect transmission and distribution networks by capturing images of assets using helicopters, drones, or ground-based teams. However, manual analysis of these images is time-consuming—an inspection flight can produce hundreds of photos per mile, and transmission lines stretch across thousands of miles nationwide. This deep learning model automates the detection of H-frame structures and their components, streamlining inspection workflows and enabling faster, more efficient analysis.