@inproceedings{a9a14f0892d94cab8b8629bec0b7f4b0,
title = "3D convolutional network based foreground feature fusion",
abstract = "With explosion of videos, action recognition has become an important research subject. This paper makes a special effort to investigate and study 3D Convolutional Network. Focused on the problem of ConvNet dependence on multiple large scale dataset, we propose a 3D ConvNet structure which incorporate the original 3D-ConvNet features and foreground 3D-ConvNet features fused by static object and motion detection. Our architecture is trained and evaluated on the standard video actions benchmarks of UCF-101 and HMDB-51, experimental results demonstrate that with merely 50\% pixels utilization, foreground ConvNet achieves satisfying performance as same as origin. With feature fusion, we achieve 83.7\% accuracy on UCF-101 exceeding original ConvNet.",
keywords = "C3D, Feature Fusion, Foreground Exaction",
author = "Hanjian Song and Lihua Tian and Chen Li",
note = "Publisher Copyright: {\textcopyright} 2018 IEEE; 20th IEEE International Symposium on Multimedia, ISM 2018 ; Conference date: 10-12-2018 Through 12-12-2018",
year = "2018",
month = jul,
day = "2",
doi = "10.1109/ISM.2018.00036",
language = "英语",
series = "Proceedings - 2018 IEEE International Symposium on Multimedia, ISM 2018",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
pages = "253--258",
booktitle = "Proceedings - 2018 IEEE International Symposium on Multimedia, ISM 2018",
}