Introduction of Fully Convolutional Networks (FCNs) has made record progress in salient object detection models. However, in order to retain the input resolutions, deconvolutional networks with unpooling are applied on top of FCNs. This will cause the increase of the computation and network model size in segmentation task. In addition, most deep learning based methods always discard effective saliency prior knowledge completely, which are shown effective. Therefore, an efficient salient object detection method based on deep learning is proposed in our work. In this model, dilated convolutions are exploited in the networks to produce the output with high resolution without pooling and adding deconvolutional networks. In this way, the parameters and depth of the network are decreased sharply compared with the traditional FCNs. Furthermore, manifold ranking model is explored for the saliency refinement to keep the spatial consistency and contour preserving. Experimental results verify that performance of our method is superior with other state-of-art methods. Meanwhile, the proposed model occupies the less model size and fastest processing speed, which is more suitable for the wearable processing systems.
Fei GUO
Xi'an University of Technology
Yuan YANG
Xi'an University of Technology
Yong GAO
Xi'an University of Technology
Ningmei YU
Xi'an University of Technology
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
Fei GUO, Yuan YANG, Yong GAO, Ningmei YU, "Efficient Salient Object Detection Model with Dilated Convolutional Networks" in IEICE TRANSACTIONS on Information,
vol. E103-D, no. 10, pp. 2199-2207, October 2020, doi: 10.1587/transinf.2019EDP7284.
Abstract: Introduction of Fully Convolutional Networks (FCNs) has made record progress in salient object detection models. However, in order to retain the input resolutions, deconvolutional networks with unpooling are applied on top of FCNs. This will cause the increase of the computation and network model size in segmentation task. In addition, most deep learning based methods always discard effective saliency prior knowledge completely, which are shown effective. Therefore, an efficient salient object detection method based on deep learning is proposed in our work. In this model, dilated convolutions are exploited in the networks to produce the output with high resolution without pooling and adding deconvolutional networks. In this way, the parameters and depth of the network are decreased sharply compared with the traditional FCNs. Furthermore, manifold ranking model is explored for the saliency refinement to keep the spatial consistency and contour preserving. Experimental results verify that performance of our method is superior with other state-of-art methods. Meanwhile, the proposed model occupies the less model size and fastest processing speed, which is more suitable for the wearable processing systems.
URL: https://global.ieice.org/en_transactions/information/10.1587/transinf.2019EDP7284/_p
Copy
@ARTICLE{e103-d_10_2199,
author={Fei GUO, Yuan YANG, Yong GAO, Ningmei YU, },
journal={IEICE TRANSACTIONS on Information},
title={Efficient Salient Object Detection Model with Dilated Convolutional Networks},
year={2020},
volume={E103-D},
number={10},
pages={2199-2207},
abstract={Introduction of Fully Convolutional Networks (FCNs) has made record progress in salient object detection models. However, in order to retain the input resolutions, deconvolutional networks with unpooling are applied on top of FCNs. This will cause the increase of the computation and network model size in segmentation task. In addition, most deep learning based methods always discard effective saliency prior knowledge completely, which are shown effective. Therefore, an efficient salient object detection method based on deep learning is proposed in our work. In this model, dilated convolutions are exploited in the networks to produce the output with high resolution without pooling and adding deconvolutional networks. In this way, the parameters and depth of the network are decreased sharply compared with the traditional FCNs. Furthermore, manifold ranking model is explored for the saliency refinement to keep the spatial consistency and contour preserving. Experimental results verify that performance of our method is superior with other state-of-art methods. Meanwhile, the proposed model occupies the less model size and fastest processing speed, which is more suitable for the wearable processing systems.},
keywords={},
doi={10.1587/transinf.2019EDP7284},
ISSN={1745-1361},
month={October},}
Copy
TY - JOUR
TI - Efficient Salient Object Detection Model with Dilated Convolutional Networks
T2 - IEICE TRANSACTIONS on Information
SP - 2199
EP - 2207
AU - Fei GUO
AU - Yuan YANG
AU - Yong GAO
AU - Ningmei YU
PY - 2020
DO - 10.1587/transinf.2019EDP7284
JO - IEICE TRANSACTIONS on Information
SN - 1745-1361
VL - E103-D
IS - 10
JA - IEICE TRANSACTIONS on Information
Y1 - October 2020
AB - Introduction of Fully Convolutional Networks (FCNs) has made record progress in salient object detection models. However, in order to retain the input resolutions, deconvolutional networks with unpooling are applied on top of FCNs. This will cause the increase of the computation and network model size in segmentation task. In addition, most deep learning based methods always discard effective saliency prior knowledge completely, which are shown effective. Therefore, an efficient salient object detection method based on deep learning is proposed in our work. In this model, dilated convolutions are exploited in the networks to produce the output with high resolution without pooling and adding deconvolutional networks. In this way, the parameters and depth of the network are decreased sharply compared with the traditional FCNs. Furthermore, manifold ranking model is explored for the saliency refinement to keep the spatial consistency and contour preserving. Experimental results verify that performance of our method is superior with other state-of-art methods. Meanwhile, the proposed model occupies the less model size and fastest processing speed, which is more suitable for the wearable processing systems.
ER -