Deep Vectorization of Technical Drawings
This addresses the need for efficient conversion of scanned technical drawings into editable vector formats, which is incremental as it builds on prior vectorization methods with new deep learning components.
The paper tackles the problem of vectorizing technical line drawings like floor plans and CAD images by introducing a method that includes deep learning-based cleaning, transformer-based vector primitive estimation, and optimization, resulting in quantitative and qualitative improvements over existing techniques.
We present a new method for vectorization of technical line drawings, such as floor plans, architectural drawings, and 2D CAD images. Our method includes (1) a deep learning-based cleaning stage to eliminate the background and imperfections in the image and fill in missing parts, (2) a transformer-based network to estimate vector primitives, and (3) optimization procedure to obtain the final primitive configurations. We train the networks on synthetic data, renderings of vector line drawings, and manually vectorized scans of line drawings. Our method quantitatively and qualitatively outperforms a number of existing techniques on a collection of representative technical drawings.