# Photo2CAD: Automated 3D solid reconstruction from 2D drawings using OpenCV

arXiv: https://arxiv.org/abs/2101.04248 · HTML: https://mlsanity.com/paper/2101.04248/

## Key facts

- Cited by 5 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 12 Jan 2021.

## Built on by (most cited first, 5 of 5)

- [Text2CAD: Text to 3D CAD Generation via Technical Drawings](https://mlsanity.com/paper/2411.06206/) (2024)
- [Drawing2CAD: Sequence-to-Sequence Learning for CAD Generation from Vector Drawings](https://mlsanity.com/paper/2508.18733/) (2025)
- [GaussianCAD: Robust Self-Supervised CAD Reconstruction from Three Orthographic Views Using 3D Gaussian Splatting](https://mlsanity.com/paper/2503.05161/) (2025)
- [Drawing-Recode: Annotation Grounding for Parametric CAD Code Generation from Raster 2D CAD Drawings](https://mlsanity.com/paper/2607.27558/) (2026)
- [DrawingsDreamer: A Unified Multi-View Engineering Drawings Generation Model](https://mlsanity.com/paper/2609.35242/) (2026)

## Abstract (arXiv)

This study showcases the utilisation of OpenCV for extracting features from photos of 2D engineering drawings. These features are then employed to reconstruct 3D CAD models in SCAD format and generate 3D point cloud data similar to LIDAR scans. Many historical mechanical, aerospace, and civil engineering designs exist only as drawings, lacking software-generated CAD or BIM models. While 2D to 3D conversion itself is not novel, the novelty of this work is in the usage of simple photos rather than scans or electronic documentation of 2D drawings. The method can also use scanned drawing data. While the approach is effective for simple shapes, it currently does not address hidden lines in CAD drawings. The Python Jupyter notebook codes developed for this purpose are accessible through GitHub.