IEEE Trans Pattern Anal Mach Intell
November 2023
Self-attention mechanisms and non-local blocks have become crucial building blocks for state-of-the-art neural architectures thanks to their unparalleled ability in capturing long-range dependencies in the input. However their cost is quadratic with the number of spatial positions hence making their use impractical in many real case applications. In this work, we analyze these methods through a polynomial lens, and we show that self-attention can be seen as a special case of a 3 rd order polynomial.
View Article and Find Full Text PDFIn this paper, a novel method for MRI volume segmentation based on region adaptive splitting and merging is proposed. The method, called Adaptive Geometric Split Merge (AGSM) segmentation, aims at finding complex geometrical shapes that consist of homogeneous geometrical 3D regions. In each volume splitting step, several splitting strategies are examined and the most appropriate is activated.
View Article and Find Full Text PDFThis article presents a novel virtual teeth drilling system designed to aid dentists, dental students, and researchers in getting acquainted with teeth anatomy, the handling of drilling instruments, and the challenges associated with drilling procedures during endodontic therapy. The system is designed to be used for educational and research purposes in dental schools. The application features a 3D face and oral cavity model constructed using anatomical data that can be adapted to the characteristics of a specific patient using either facial photographs or 3D data.
View Article and Find Full Text PDF