作者:Narayanan Sundaram 作者单位:Electrical Engineering and Computer Sciences University of California at Berkeley 加工时间:2013-11-16 信息来源:EECS 索取原文[160 页]
关键词:计算机视觉;机器学习算法;数值优化 摘 要:In this thesis, we describe the computational characteristics of computer vision work-loads using patterns. We show examples of how careful numerical optimization has led toincreased speedups on many computer vision and machine learning algorithms includingsupport vector machines, optical ow, point tracking, image contour detection and video segmentation.