A Hybrid Transfer Learning Mechanism for Object Classification across View

CASIA OpenIR > 智能感知与计算研究中心

	A Hybrid Transfer Learning Mechanism for Object Classification across View
	Yi Mo; Zhaoxiang Zhang; Yunhong Wang
	2012-12-12
会议名称	11th International Conference on Machine Learning and Applications
会议录名称	ICMLA 2012
会议日期	12-15 December 2012
会议地点	Boca Raton, Florida, USA
摘要	Object classification in traffic scene is of vital importance to intelligent traffic surveillance. In real applications, the shooting view changes frequently in different scenes, which leads to sharp accuracy decrease since source and target domain samples do not follow the same distribution anymore. On the other hand, manual labeling training samples is time and labor consuming. Transfer learning approaches are to utilize the knowledge learnt from source view for target object classification. In this paper, we propose a hybrid transfer learning mechanism combining two single transfer approaches to gap the divergence of different domain distributions. An instance-based transfer approach is implemented to label target samples that represent target domain distribution best. And a feature-based transfer framework is to learn a strong classifier for target domain with both labeled source and target domain samples. Experimental results indicate that our approach outperforms traditional machine learning and single transfer learning methods.
关键词	Transfer Learning Traffic Scene Surveillance Object Classification
文献类型	会议论文
条目标识符	http://ir.ia.ac.cn/handle/173211/13254
专题	智能感知与计算研究中心
通讯作者	Zhaoxiang Zhang
推荐引用方式 GB/T 7714	Yi Mo,Zhaoxiang Zhang,Yunhong Wang. A Hybrid Transfer Learning Mechanism for Object Classification across View[C],2012.