Active learning for transferrable object classification in cross-view traffic scene surveillance

Authors:
Zhaoxiang Zhang;Jun Tang;Yuhang Zhao;Yunhong Wang;Jianyun Liu
Affiliations:
Laboratory of Intelligent Recognition and Image Processing, Beijing Key Laboratory of Digital Media, Beihang University, Beijing, China;Laboratory of Intelligent Recognition and Image Processing, Beijing Key Laboratory of Digital Media, Beihang University, Beijing, China;Laboratory of Intelligent Recognition and Image Processing, Beijing Key Laboratory of Digital Media, Beihang University, Beijing, China;Laboratory of Intelligent Recognition and Image Processing, Beijing Key Laboratory of Digital Media, Beihang University, Beijing, China;Laboratory of Intelligent Recognition and Image Processing, Beijing Key Laboratory of Digital Media, Beihang University, Beijing, China
Venue:
PCM'12 Proceedings of the 13th Pacific-Rim conference on Advances in Multimedia Information Processing
Year:
2012

Citing 5
Cited 0

Learning to Classify Text Using Support Vector Machines: Methods, Theory and Algorithms

Learning to Classify Text Using Support Vector Machines: Methods, Theory and Algorithms
View independent vehicle/person classification

Proceedings of the ACM 2nd international workshop on Video surveillance & sensor networks
Cast Shadow Removal with GMM for Surface Reflectance Component

ICPR '06 Proceedings of the 18th International Conference on Pattern Recognition - Volume 01
Domain adaptation with structural correspondence learning

EMNLP '06 Proceedings of the 2006 Conference on Empirical Methods in Natural Language Processing
Evaluating Color Descriptors for Object and Scene Recognition

IEEE Transactions on Pattern Analysis and Machine Intelligence

Quantified Score

Hi-index	0.00

Visualization

Abstract

We discuss the problem of object classification in cross-view traffic scene surveillance videos in this paper. To classify moving objects in traffic scene videos into pedestrian, bicycle and variety of vehicles, an effective intelligent classification framework has been proposed which takes advantage of a transfer machine learning method to bridge the gap between source scene data and target scene data. The transfer learning algorithm makes one classifier adaptive to perspective changes instead of training two different classifiers for corresponding perspectives. The samples transferred from source scene database have saved much manual labeling work on target scene database. In this paper, we propose an active transfer learning method to decrease manual labeling work further for target scene traffic object classification. Redundant experiments are conducted and experimental results demonstrate the effectiveness and convenience of our approach.