A note on retrieval of visually similar distress regions in subway tunnel images Introduction of deep features extracted by semantic segmentation network

摘要

Automatic retrieval of visually similar distress regions in subway tunnel images can facilitate the maintenance of subway tunnels and help with the analysis of distress regions. In this paper, we construct an automatic retrieval method for visually similar distress regions in subway tunnel images. The method receives query distress images and gives retrieval results from the candidate set. The retrieval is performea by comparing the features extracted by deep convolutional neural networks. As a result, the experimental results confirm that the retrieved images involve regions semantically and spatially similar to the query distress regions.