Abstract & Overview
"Railway semantic segmentation is the task of highlighting rail blades in images taken from the ego-view of the train. Solving this task allows for further image processing on the rails, which can be used for more complex problems such as switch or fault detection. In this paper we approach the railway semantic segmentation using two deep architectures from the U-Net family, U-Net and ResUNet++, using the most comprehensive dataset available at the time of writing from the railway scene, namely RailSem19. We also investigate the effects of image augmentations and different training dataset sizes, as well as the performance of the models on dark images. We have compared our solution to other approaches and obtained competitive results with larger scores. Keywords and phrases: Binary Semantic Segmentation, Encoder-decoder, Railway blades, Deep Learning. "