We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

cs.CV

Change to browse by:

References & Citations

DBLP - CS Bibliography

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo

Computer Science > Computer Vision and Pattern Recognition

Title: FastGeodis: Fast Generalised Geodesic Distance Transform

Abstract: The FastGeodis package provides an efficient implementation for computing Geodesic and Euclidean distance transforms (or a mixture of both) targeting efficient utilisation of CPU and GPU hardwares. In particular, it implements paralellisable raster scan method from Criminisi et al, where elements in row (2D) or plane (3D) can be computed with parallel threads. This package is able to handle 2D as well as 3D data where it achieves up to 15x speed-up on CPU and up to 60x speed-up on GPU as compared to existing open-source libraries, which uses non-parallelisable single-thread CPU implementation. The performance speed-ups reported here were evaluated using 3D volume data on Nvidia GeForce Titan X (12 GB) with 6-Core Intel Xeon E5-1650 CPU. This package is available at: this https URL
Comments: Under-review at Journal of Open Source Software (JOSS)
Subjects: Computer Vision and Pattern Recognition (cs.CV); Image and Video Processing (eess.IV)
Cite as: arXiv:2208.00001 [cs.CV]
  (or arXiv:2208.00001v1 [cs.CV] for this version)

Submission history

From: Muhammad Asad [view email]
[v1] Tue, 26 Jul 2022 15:01:37 GMT (169kb,D)
[v2] Wed, 23 Nov 2022 23:33:37 GMT (216kb,D)

Link back to: arXiv, form interface, contact.