Computer Vision (CV) has the potential to tremendously reduce costs and increase the efficiency of preventative maintenance and inspection. In particular, the recognition and identification of structural damage with automated systems would reduce or eliminate the need for a human inspector. CV not only reduces Operational & Maintenance (O&M) costs, but also introduces the possibility of dam­age detection on physically inaccessible locations. Mining equipment, for example, is often difficult for human inspectors to assess. The authors propose a Convolutional Neural Network (CNN) based method­ology for the recognition and identification of the presence and type of damage. A CNN is a deep feed-forward Artificial Neural Network that includes convolutional and pooling layers. Conceptually rooted in a human’s visual cortex, CNN’s are invariant to image scale, surface type, and damage location. The proposed methodology is validated on a synthetic data set and crack damage recognition is demonstrated on real concrete bridge crack images.