CUHOG: A HOG implementation in C# on CUDA, which is compatible with Pedro Felzenszwalb's 31-dim HOG seamlessly. CUDA. C#

Translated title of the contribution: CUHOG: A HOG implementation in C# on CUDA, which is compatible with Pedro Felzenszwalb's 31-dim HOG seamlessly.: CUDA. C#

Pedersoli Marco, Gonzalez Jordi, Hu Xu, Roca Xavier

Research output: Non-textual formSoftwareResearch

Abstract

This implementation is based on fastHOG [1] and groundHOG [2]. Our real-time object detection system will be released in next version associated with the Coarse-to-fine [3]. An example with matlab interface will also give. Currently, for a 640x480 image takes only 50 ms in average and the real-time object detection can achieve 10 FPS in GTX560 and 2.6GHZ 4-core CPU.
Translated title of the contributionCUHOG: A HOG implementation in C# on CUDA, which is compatible with Pedro Felzenszwalb's 31-dim HOG seamlessly.: CUDA. C#
Original languageUndefined/Unknown
Publication statusPublished - 30 Sep 2013

Fingerprint Dive into the research topics of 'CUHOG: A HOG implementation in C# on CUDA, which is compatible with Pedro Felzenszwalb's 31-dim HOG seamlessly. CUDA. C#'. Together they form a unique fingerprint.

  • Cite this