Object tracking by using online learning with deep neural network features깊은 신경망 특징을 이용한 실시간 학습에 의한 물체 추적알고리즘

Cited 0 time in webofscience Cited 0 time in scopus
  • Hit : 634
  • Download : 0
The deep learning of neural network works on vision recognition and classification tasks briskly, and it can extract great features of an image for classification recently. Therefore, many approaches in the field of visual object tracking studied with this advance in two-ways with these characteristics. First, they can regard tracking problem as classifying each video and frame by learning all dataset. Second, use the deep neural network as feature generator and use other classifiers for using their features such as Support Vector Machine(SVM). On the second part, the features can be used to learn discriminative target appearance models like online SVM. We propose an adaptive visual tracking framework based on the state-of-the-art ferns binary feature detection method with Convolution Neural Network(CNN) features. The ferns feature detection obtaining simple binary feature and it can make online training samples for the classifier. Our framework makes binary feature consist of ferns features and CNN features which are learning classifier online to provide adaptive tracking. These features are simple and powerful features for learning classifier of the detector and make it robust to target object with time. We test our proposed method with state-of-the-art trackers on existing tracking benchmarks.
Advisors
Kim, Dae-Shikresearcher김대식researcher
Description
한국과학기술원 :전기및전자공학부,
Publisher
한국과학기술원
Issue Date
2016
Identifier
325007
Language
eng
Description

학위논문(석사) - 한국과학기술원 : 전기및전자공학부, 2016.8 ,[v, 36 p. :]

Keywords

Object tracking; deep learning; neural network; binary feature; online learning; 물체 추적; 깊은 학습; 인공신경망; 이진 특징; 실시간 학습

URI
http://hdl.handle.net/10203/221670
Link
http://library.kaist.ac.kr/search/detail/view.do?bibCtrlNo=663464&flag=dissertation
Appears in Collection
EE-Theses_Master(석사논문)
Files in This Item
There are no files associated with this item.

qr_code

  • mendeley

    citeulike


rss_1.0 rss_2.0 atom_1.0