Abstract: The paper introduces a generic deep learning–based method using a sliding window computing algorithm based on long short-term memory (LSTM) networks for the classification of potential anomalies (e.g., cracks, potholes, ...