In this paper, we introduce a fully automatic approach to construct action datasets from noisy Web video search results. The idea is based on combining cluster structure analysis and density-based outlier detection. For a specific action concept, first, we download its Web top search videos and segment them into video shots. We then organize these shots into subsets using density-based hierarchy clustering. For each set, we rank its shots by their outlier degrees which are determined as their isolatedness with respect to their surroundings. Finally, we collect high ranked shots as training data for the action concept. We demonstrate that with action models trained by our data, we can obtain promising precision rates in the task of action classification while offering the advantage of fully automatic, scalable learning. Experiment results on UCF11, a challenging action dataset, show the effectiveness of our method.
Nga Hang DO
The University of Electro-Communications
Keiji YANAI
The University of Electro-Communications
The copyright of the original papers published on this site belongs to IEICE. Unauthorized use of the original or translated papers is prohibited. See IEICE Provisions on Copyright for details.
Copy
Nga Hang DO, Keiji YANAI, "Automatic Retrieval of Action Video Shots from the Web Using Density-Based Cluster Analysis and Outlier Detection" in IEICE TRANSACTIONS on Information,
vol. E99-D, no. 11, pp. 2788-2795, November 2016, doi: 10.1587/transinf.2016EDP7108.
Abstract: In this paper, we introduce a fully automatic approach to construct action datasets from noisy Web video search results. The idea is based on combining cluster structure analysis and density-based outlier detection. For a specific action concept, first, we download its Web top search videos and segment them into video shots. We then organize these shots into subsets using density-based hierarchy clustering. For each set, we rank its shots by their outlier degrees which are determined as their isolatedness with respect to their surroundings. Finally, we collect high ranked shots as training data for the action concept. We demonstrate that with action models trained by our data, we can obtain promising precision rates in the task of action classification while offering the advantage of fully automatic, scalable learning. Experiment results on UCF11, a challenging action dataset, show the effectiveness of our method.
URL: https://global.ieice.org/en_transactions/information/10.1587/transinf.2016EDP7108/_p
Copy
@ARTICLE{e99-d_11_2788,
author={Nga Hang DO, Keiji YANAI, },
journal={IEICE TRANSACTIONS on Information},
title={Automatic Retrieval of Action Video Shots from the Web Using Density-Based Cluster Analysis and Outlier Detection},
year={2016},
volume={E99-D},
number={11},
pages={2788-2795},
abstract={In this paper, we introduce a fully automatic approach to construct action datasets from noisy Web video search results. The idea is based on combining cluster structure analysis and density-based outlier detection. For a specific action concept, first, we download its Web top search videos and segment them into video shots. We then organize these shots into subsets using density-based hierarchy clustering. For each set, we rank its shots by their outlier degrees which are determined as their isolatedness with respect to their surroundings. Finally, we collect high ranked shots as training data for the action concept. We demonstrate that with action models trained by our data, we can obtain promising precision rates in the task of action classification while offering the advantage of fully automatic, scalable learning. Experiment results on UCF11, a challenging action dataset, show the effectiveness of our method.},
keywords={},
doi={10.1587/transinf.2016EDP7108},
ISSN={1745-1361},
month={November},}
Copy
TY - JOUR
TI - Automatic Retrieval of Action Video Shots from the Web Using Density-Based Cluster Analysis and Outlier Detection
T2 - IEICE TRANSACTIONS on Information
SP - 2788
EP - 2795
AU - Nga Hang DO
AU - Keiji YANAI
PY - 2016
DO - 10.1587/transinf.2016EDP7108
JO - IEICE TRANSACTIONS on Information
SN - 1745-1361
VL - E99-D
IS - 11
JA - IEICE TRANSACTIONS on Information
Y1 - November 2016
AB - In this paper, we introduce a fully automatic approach to construct action datasets from noisy Web video search results. The idea is based on combining cluster structure analysis and density-based outlier detection. For a specific action concept, first, we download its Web top search videos and segment them into video shots. We then organize these shots into subsets using density-based hierarchy clustering. For each set, we rank its shots by their outlier degrees which are determined as their isolatedness with respect to their surroundings. Finally, we collect high ranked shots as training data for the action concept. We demonstrate that with action models trained by our data, we can obtain promising precision rates in the task of action classification while offering the advantage of fully automatic, scalable learning. Experiment results on UCF11, a challenging action dataset, show the effectiveness of our method.
ER -