EDRS: an Automatic System to Recognize Electrical Drawings

Electrical drawings, consisting of electrical symbol elements, connection lines and text annotations, describe the layout of electrical equipment. However, power system technicians often have to manually redraw them from image format to digital files, which is challenging. Here we propose EDRS—a deep learning based automatic Electrical Drawings Recognition System for electrical drawings. Specifically, we first develop a Faster- RCNN based model to accomplish the subtask of electrical symbols automatic recognition. Then we provide a text detection and recognition model to recognize the content of electrical text annotations, while the electrical connection lines are also recognized by a pixel-level digital image processing approach. Finally, we carry out an association analysis on electrical elements, with which we identify the topological relationship of electrical elements and group the electrical elements for determining their types. Extensive experiments on several real-world power systems demonstrate that our EDRS can effectively identify the electrical elements and their relationship in drawings.