인공지능을 활용한 3차원 형상 인식 실증적 연구 (Ⅱ)

초록

With AI advancements, interest in 3D shape recognition is growing. This study aims to automate feature recognition in aircraft component manufacturing using CAD data and the SsdNet model, targeting materials over 3 meters, unlike standard features from prior research. Voxel transformation and Overlapping Segmentation optimized performance, achieving 83.3% accuracy across eight features. However, complex curves and overlaps reduced recognition rates, indicating a need for further research. It is expected to enhance productivity in smart manufacturing by automating feature recognition in NC programming.

키워드