Online Tracking by Learning Discriminative Saliency Map with Convolutional Neural Network

Seunghoon Hong, Tackgeun You, Suha Kwak, Bohyung Han
Proceedings of the 32nd International Conference on Machine Learning, PMLR 37:597-606, 2015.

Abstract

We propose an online visual tracking algorithm by learning discriminative saliency map using Convolutional Neural Network (CNN). Given a CNN pre-trained on a large-scale image repository in offline, our algorithm takes outputs from hidden layers of the network as feature descriptors since they show excellent representation performance in various general visual recognition problems. The features are used to learn discriminative target appearance models using an online Support Vector Machine (SVM). In addition, we construct target-specific saliency map by back-projecting CNN features with guidance of the SVM, and obtain the final tracking result in each frame based on the appearance model generatively constructed with the saliency map. Since the saliency map reveals spatial configuration of target effectively, it improves target localization accuracy and enables us to achieve pixel-level target segmentation. We verify the effectiveness of our tracking algorithm through extensive experiment on a challenging benchmark, where our method illustrates outstanding performance compared to the state-of-the-art tracking algorithms.

Cite this Paper

BibTeX
@InProceedings{pmlr-v37-hong15, title = {Online Tracking by Learning Discriminative Saliency Map with Convolutional Neural Network}, author = {Hong, Seunghoon and You, Tackgeun and Kwak, Suha and Han, Bohyung}, booktitle = {Proceedings of the 32nd International Conference on Machine Learning}, pages = {597--606}, year = {2015}, editor = {Bach, Francis and Blei, David}, volume = {37}, series = {Proceedings of Machine Learning Research}, address = {Lille, France}, month = {07--09 Jul}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v37/hong15.pdf}, url = {https://proceedings.mlr.press/v37/hong15.html}, abstract = {We propose an online visual tracking algorithm by learning discriminative saliency map using Convolutional Neural Network (CNN). Given a CNN pre-trained on a large-scale image repository in offline, our algorithm takes outputs from hidden layers of the network as feature descriptors since they show excellent representation performance in various general visual recognition problems. The features are used to learn discriminative target appearance models using an online Support Vector Machine (SVM). In addition, we construct target-specific saliency map by back-projecting CNN features with guidance of the SVM, and obtain the final tracking result in each frame based on the appearance model generatively constructed with the saliency map. Since the saliency map reveals spatial configuration of target effectively, it improves target localization accuracy and enables us to achieve pixel-level target segmentation. We verify the effectiveness of our tracking algorithm through extensive experiment on a challenging benchmark, where our method illustrates outstanding performance compared to the state-of-the-art tracking algorithms.} }
Endnote
%0 Conference Paper %T Online Tracking by Learning Discriminative Saliency Map with Convolutional Neural Network %A Seunghoon Hong %A Tackgeun You %A Suha Kwak %A Bohyung Han %B Proceedings of the 32nd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2015 %E Francis Bach %E David Blei %F pmlr-v37-hong15 %I PMLR %P 597--606 %U https://proceedings.mlr.press/v37/hong15.html %V 37 %X We propose an online visual tracking algorithm by learning discriminative saliency map using Convolutional Neural Network (CNN). Given a CNN pre-trained on a large-scale image repository in offline, our algorithm takes outputs from hidden layers of the network as feature descriptors since they show excellent representation performance in various general visual recognition problems. The features are used to learn discriminative target appearance models using an online Support Vector Machine (SVM). In addition, we construct target-specific saliency map by back-projecting CNN features with guidance of the SVM, and obtain the final tracking result in each frame based on the appearance model generatively constructed with the saliency map. Since the saliency map reveals spatial configuration of target effectively, it improves target localization accuracy and enables us to achieve pixel-level target segmentation. We verify the effectiveness of our tracking algorithm through extensive experiment on a challenging benchmark, where our method illustrates outstanding performance compared to the state-of-the-art tracking algorithms.
RIS
TY - CPAPER TI - Online Tracking by Learning Discriminative Saliency Map with Convolutional Neural Network AU - Seunghoon Hong AU - Tackgeun You AU - Suha Kwak AU - Bohyung Han BT - Proceedings of the 32nd International Conference on Machine Learning DA - 2015/06/01 ED - Francis Bach ED - David Blei ID - pmlr-v37-hong15 PB - PMLR DP - Proceedings of Machine Learning Research VL - 37 SP - 597 EP - 606 L1 - http://proceedings.mlr.press/v37/hong15.pdf UR - https://proceedings.mlr.press/v37/hong15.html AB - We propose an online visual tracking algorithm by learning discriminative saliency map using Convolutional Neural Network (CNN). Given a CNN pre-trained on a large-scale image repository in offline, our algorithm takes outputs from hidden layers of the network as feature descriptors since they show excellent representation performance in various general visual recognition problems. The features are used to learn discriminative target appearance models using an online Support Vector Machine (SVM). In addition, we construct target-specific saliency map by back-projecting CNN features with guidance of the SVM, and obtain the final tracking result in each frame based on the appearance model generatively constructed with the saliency map. Since the saliency map reveals spatial configuration of target effectively, it improves target localization accuracy and enables us to achieve pixel-level target segmentation. We verify the effectiveness of our tracking algorithm through extensive experiment on a challenging benchmark, where our method illustrates outstanding performance compared to the state-of-the-art tracking algorithms. ER -
APA
Hong, S., You, T., Kwak, S. & Han, B.. (2015). Online Tracking by Learning Discriminative Saliency Map with Convolutional Neural Network. Proceedings of the 32nd International Conference on Machine Learning, in Proceedings of Machine Learning Research 37:597-606 Available from https://proceedings.mlr.press/v37/hong15.html.

Related Material