基于在线多实例学习的跟踪研究

摘要

介绍一种基于在线多实例学习的目标跟踪方法，在线多实例学习在目标跟踪中的应用是采用基于多实例学习的自适应表观模型，解决了训练样本中存在的模糊歧义问题，探讨了处理训练正包中正样本和负样本的问题，指出了今后多实例学习目标跟踪的研究方向．%A online multi-instance learning-based object tracking approach is presented, multi-instance learning-based adaptive appear model is used by object tracking , the ambiguity of train samples is solved, the problem of positive bag with positive samples and negative samples is pointed, the future study of multi-instance learning-based object tracking is provided.