Supervisor: Gian Antonio Susto
Creation Date: 05/10/2025 15:48
The goal of this thesis is to develop a system capable of digitizing freehand technical sketches that include geometric shapes and dimensional annotations. The system should automatically recognize the geometric entities (lines, circles, arcs, etc.) and their corresponding dimensions and constraints, generating a parametric digital model equivalent to the original drawing.
The ultimate objective is to produce standard CAD files (e.g., DXF or STEP) directly from the hand-drawn sketch, thereby simplifying the transition from manual concept design to digital modeling — an important step toward fully digitalized design workflows within Breton’s product development process.
Dataset type: Data to be acquired
Dataset description: The data used in this study consist of images of hand-drawn technical sketches that include both geometric elements and dimensional annotations. These sketches simulate early-stage engineering drawings, where designers communicate shape, proportion, and key measurements before creating a digital CAD model. The dataset was created to capture a wide variety of drawing styles, levels of complexity, and degrees of precision, ensuring that the recognition algorithms can generalize to real industrial conditions.
List of Methods: The process integrates image processing, computer vision, optical character recognition (OCR), and geometric reconstruction techniques.
Machine Learning, Deep Learning