During the long-term operation of hydro-junction infrastructure, water flow erosion causes concrete surfaces to crack, resulting in seepage, spalling, and rebar exposure. To ensure infrastructure safety, detecting such damage is critical. We propose a highly accurate damage detection method using a deep convolutional neural network with transfer learning. First, we collected images from hydro-junction infrastructure using a high-definition camera. Second, we preprocessed the images using an image expansion method. Finally, we modified the structure of Inception-v3 and trained the network using transfer learning to detect damage. The experiments show that the accuracy of the proposed damage detection method is 96.8%, considerably higher than the accuracy of a support vector machine. The results demonstrate that our damage detection method achieves better damage detection performance.

Crack detection on dam surfaces is an important task for safe inspection of hydropower stations. More and more object detection methods based on deep learning are being applied to crack detection. However, most of the methods can only achieve the classification and rough location of cracks. Pixel-level crack detection can provide more intuitive and accurate detection results for dam health assessment. To realize pixel-level crack detection, a method of crack detection on dam surface (CDDS) using deep convolution network is proposed. First, we use an unmanned aerial vehicle (UAV) to collect dam surface images along a predetermined trajectory. Second, raw images are cropped. Then crack regions are manually labelled on cropped images to create the crack dataset, and the architecture of CDDS network is designed. Finally, the CDDS network is trained, validated and tested using the crack dataset. To validate the performance of the CDDS network, the predicted results are compared with ResNet152-based, SegNet, UNet and fully convolutional network (FCN). In terms of crack segmentation, the recall, precision, F-measure and IoU are 80.45%, 80.31%, 79.16%, and 66.76%. The results on test dataset show that the CDDS network has better performance for crack detection of dam surfaces.