首页>
外国专利>
DOMAIN ADAPTATION FOR IMAGE CLASSIFICATION USING CLASS PRIOR PROBABILITY
DOMAIN ADAPTATION FOR IMAGE CLASSIFICATION USING CLASS PRIOR PROBABILITY
展开▼
机译：使用类别先验概率进行图像分类的域自适应
展开▼
页面导航
摘要
著录项
相似文献
摘要
PROBLEM TO BE SOLVED: To provide a domain adaptation method for image classification using a class prior probability.;SOLUTION: In camera-based object labeling, a boost classifier f is trained so as to classify an image represented by a feature vector x using a target domain training set DT of labeled feature vectors representing images acquired by the same camera and a plurality of source domain training sets DS1, ..., DSN acquired by other cameras. The training applies an adaptive boosting (AdaBoost) algorithm to generate a basic classifier hr(x) and weight βr. An r-th iteration of the AdaBoost algorithm trains basic classifier candidates hkr(x) each trained on a training set DT(Union)DSk, and selects the hr(x) from preliminarily trained basic classifier candidates. The target domain training set DT may be expanded based on a prior estimate of the label distribution for the target domain.;SELECTED DRAWING: None;COPYRIGHT: (C)2016,JPO&INPIT
展开▼
