EFFECTIVENESS OF OBJECT RECOGNITION ALGORITHMS FOR MODELING AIRCRAFT AND UAV STRUCTURAL COMPONENTS IN AUTOCAD PROJECTS

Computer-aided design environments increasingly require automated analysis tools capable of interpreting complex technical drawings. This study examines the effectiveness of object recognition algorithms applied to AutoCAD projects, with a particular focus on aircraft and unmanned aerial vehicle structural modeling. A structured processing framework is proposed, including vector-to-raster conversion, adaptive binarization, morphological preprocessing, edge extraction, contour analysis, and rule-based classification. The mathematical formulation of each stage is presented to ensure deterministic and interpretable behavior. Experimental evaluation was carried out on a mixed dataset comprising general engineering drawings and aerospace-related structural layouts. Performance was assessed using precision, recall, F1-score, intersection-over-union, and processing time. The results show that the proposed approach maintains high detection accuracy and stable performance across varying drawing densities and geometric complexities, while remaining computationally feasible for standard engineering workstations. These findings demonstrate the potential of geometry-driven object recognition methods to support automated analysis in CAD-based engineering workflows.