U-Net-based electrical secondary drawing character information identification method and system

The invention discloses a U-Net-based electrical secondary drawing character information identification method and system, and the method comprises the steps: collecting a historical electrical secondary drawing, and carrying out the slicing of an image; using the processed image to train a U-Net deep learning model; and performing character detection on the input electrical secondary drawing by using the trained model. Compared with a traditional manual method or a simple image processing technology, character detection and recognition are conducted on the electrical secondary drawing through the U-Net deep learning model, character information in the drawing can be recognized more accurately, and meanwhile the automatic processing degree is greatly improved. And the working efficiency is improved. And adaptability to complex layout and information intensity in an electrical secondary drawing is realized. And even under the condition of excessive associated parts or dense drawing information, the character i