Ortego, DiegoORCID: 0000-0002-1011-3610, McGuinness, KevinORCID: 0000-0003-1336-6477, SanMiguel, Juan C., Arazo, Eric, Martínez, José M. and O'Connor, Noel E.ORCID: 0000-0002-4033-9135
(2019)
On guiding video object segmentation.
In: International Conference on Content-Based Multimedia Indexing, 4-6 Sept 2019, Dublin, Ireland.
This paper presents a novel approach for segmenting moving objects in unconstrained environments using guided convolutional neural networks. This guiding process relies on foreground masks from independent algorithms (i.e. state-of-the-art algorithms) to implement an attention mechanism that incorporates the spatial location of foreground and background to compute their separated representations. Our approach initially extracts two kinds of features for each frame using colour and optical flow information. Such features are combined following
a multiplicative scheme to benefit from their complementarity. These unified colour and motion features are later processed to obtain the separated foreground and background representations. Then, both independent representations are concatenated and decoded to perform foreground segmentation. Experiments
conducted on the challenging DAVIS 2016 dataset demonstrate that our guided representations not only outperform non-guided, but also recent and top-performing video object segmentation algorithms.
Metadata
Item Type:
Conference or Workshop Item (Paper)
Event Type:
Conference
Refereed:
Yes
Uncontrolled Keywords:
Video object segmentation; foreground segmentation; attention; deep learning
Spanish Government (MobiNetVideo TEC2017-88169-R), UAM-BANCO SANTANDER ´ con Europa (Red Yerun)” (2017/YERUN/02 (SOFDL), Science Foundation Ireland (SFI/12/RC/2289 and SFI/15/SIRG/3283)
ID Code:
23798
Deposited On:
02 Oct 2019 15:09 by
Diego Ortego Hernández
. Last Modified 27 Oct 2021 12:25