Incremental Training of SVM-based Human Detector

被引：1

作者：

Hanyu, Tatsuya ^{[1
]}

Zhao, Qiangfu ^{[1
]}

机构：

[1] Univ Aizu, Dept Comp Sci, Fukushima, Japan

来源：

2017 IEEE 11TH INTERNATIONAL SYMPOSIUM ON EMBEDDED MULTICORE/MANY-CORE SYSTEMS-ON-CHIP (MCSOC 2017) | 2017年

关键词：

human detection; support vector machine; incremental learning;

D O I：

10.1109/MCSoC.2017.25

中图分类号：

TP3 [计算技术、计算机技术];

学科分类号：

0812 ;

摘要：

To design a good human detector, we may collect a huge number of data, and train the detector off-line. However, even if the training data set is very large, it may not contain enough information for some particular environment, and the obtained model may not work well. In this paper, we study incremental learning of a support vector machine-based human detector in an office environment, and investigate the "growth process" of the detector. Experimental results show that it is possible to obtain a good human detector customized to a certain environment with less data via incremental learning.

引用

页码：181 / 185

页数：5