深層学習による配管画像認識手法の評価

抄録

Regular maintenance and repair of piping systems are essential in factories. Because of the missing of design layout of piping systems, people must make an observation previously. However, piping systems are too complex for human to check the details and checking parts of piping is a time-consuming process. Therefore, this study aims to develop a deep learning-based approach to recognize the parts of piping systems in factories automatically. In this paper, we trained and evaluated the models based on object detection and segment segmentation methods using transfer learning techniques. The experimental result showed the performance of the method.