Geometric Primitive

Geometric primitive research focuses on representing complex shapes using simpler, fundamental building blocks like lines, circles, and superquadrics. Current research emphasizes developing robust and efficient algorithms, often leveraging deep learning architectures such as transformers and convolutional neural networks, to segment and recognize these primitives in various data modalities, including images and point clouds. This work is significant for advancing fields like computer vision, robotics, and digital humanities, enabling tasks such as object recognition, scene understanding, and historical document analysis. The development of efficient and accurate primitive-based representations is crucial for improving the performance and generalizability of many computer vision and robotics applications.