Automatická konverzia naskenovanej projektovej dokumentácie do strojovo čitateľnej podoby

Abstract

The main goal of this diploma thesis was to design and implement a software solution enabling automatic digitization of electrical technical drawings. The resulting application is divided into several functional parts, each of which plays an important role in the processing workflow – from image preprocessing, through detection of individual components, texts, and connections, to their interpretation and structured storage. The thesis describes the use of modern methods of computer vision and machine learning, including their advantages and limitations. It also includes a compar- ison of different approaches for detecting and recognizing data from technical drawings, as well as the design and implementation of various solution variants. The final application serves as a tool for efficient digitization of drawing documentation, which can subsequently be further analyzed or visualized.