摘要
文章利用一个二值分类器、P-N分类器来改善未标记的数据的分类。P-N分类器需要利用数据集进行模型的学习,学习过程是通过正样本和负样本的约束,P-N分类器被用于目标检测以及追踪的在线学习。试验表明,利用P-N分类器的学习算法能够从一段未标记的视频序列检测、追踪指定目标。
This paper discusses the improvement on the processing of structured unlabeled data with the binary classifier and the P-N classifier.The P-N classifier,used in the online learning of target detection and tracking,needs the data for the learning of trained model whose process is constrained by the positive and negative samples.The experiment shows that the learning algorithm of the P-N classifier is capable of detecting and tracking the designated object in an unlabeled segment video sequence.
出处
《宁波工程学院学报》
2012年第1期52-54,共3页
Journal of Ningbo University of Technology
基金
2011年宁波工程学院学生科研项目
关键词
目标追踪
目标检测
P-N学习算法
target tracking
target detection
P-N learning algorithm