<p>Aircraft corrosion is a significant issue for the aerospace sector as it compromises the structural integrity, safety of passengers, and performance of an aircraft during the lifecycle of the aircraft. The common inspection methods that are used include manual visual inspection and Non-Destructive Testing (NDT), which are always subjective, time-consuming, and fail to detect any damage in a structure, especially in its complicated structure. Such constraints have increased the demand for automated and intelligent systems of inspection. Deep learning (DL) has become a disruptive solution to corrosion assessment because of its ability to extract features, segment semantics, and predict assessment through high-dimensional image data. The presented review summarizes the progress in the state-of-the-art DL architecture used to detect aircraft corrosion, such as Convolutional Neural Networks (CNNs), Fully Convolutional Networks (FCNs), U-Net variants, Vision Transformers (ViTs), and ensemble models. The research articles examined have shown better results in relation to traditional image processing techniques, with the assessment criteria being accuracy, precision, recall, F1-score, and mean Intersection over Union (mIoU). The major enabling strategies that are used to address the lack of data and environmental variability are also examined, such as advanced data augmentation, transfer learning, and the construction of corrosion-specific annotated datasets. Moreover, innovations including multi-modal sensing, Explainable Artificial Intelligence (XAI), federated and edge cloud computing are used in real-life operation conditions with enhanced transparency and real-time performance.</p>