The aim of this paper is to propose effective attentional regions for fine-grained visual recognition. Based on the Spatial Transformers' capability of spatial manipulation within networks, we propose an extension model, the Attention-Guided Spatial Transformer Networks (AG-STNs). This model can guide the Spatial Transformers with hard-coded attentional regions at first. Then such guidance can be turned off, and the network model will adjust the region learning in terms of the location and scale. Such adjustment is conditioned to the classification loss so that it is actually optimized for better recognition results. With this model, we are able to successfully capture detailed attentional information. Also, the AG-STNs are able to capture attentional information in multiple levels, and different levels of attentional information are complementary to each other in our experiments. A fusion of them brings better results.
Dichao LIU
Nagoya University
Yu WANG
Ritsumeikan University
Jien KATO
Ritsumeikan University
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
Dichao LIU, Yu WANG, Jien KATO, "Attention-Guided Spatial Transformer Networks for Fine-Grained Visual Recognition" in IEICE TRANSACTIONS on Information,
vol. E102-D, no. 12, pp. 2577-2586, December 2019, doi: 10.1587/transinf.2019EDP7045.
Abstract: The aim of this paper is to propose effective attentional regions for fine-grained visual recognition. Based on the Spatial Transformers' capability of spatial manipulation within networks, we propose an extension model, the Attention-Guided Spatial Transformer Networks (AG-STNs). This model can guide the Spatial Transformers with hard-coded attentional regions at first. Then such guidance can be turned off, and the network model will adjust the region learning in terms of the location and scale. Such adjustment is conditioned to the classification loss so that it is actually optimized for better recognition results. With this model, we are able to successfully capture detailed attentional information. Also, the AG-STNs are able to capture attentional information in multiple levels, and different levels of attentional information are complementary to each other in our experiments. A fusion of them brings better results.
URL: https://global.ieice.org/en_transactions/information/10.1587/transinf.2019EDP7045/_p
Copy
@ARTICLE{e102-d_12_2577,
author={Dichao LIU, Yu WANG, Jien KATO, },
journal={IEICE TRANSACTIONS on Information},
title={Attention-Guided Spatial Transformer Networks for Fine-Grained Visual Recognition},
year={2019},
volume={E102-D},
number={12},
pages={2577-2586},
abstract={The aim of this paper is to propose effective attentional regions for fine-grained visual recognition. Based on the Spatial Transformers' capability of spatial manipulation within networks, we propose an extension model, the Attention-Guided Spatial Transformer Networks (AG-STNs). This model can guide the Spatial Transformers with hard-coded attentional regions at first. Then such guidance can be turned off, and the network model will adjust the region learning in terms of the location and scale. Such adjustment is conditioned to the classification loss so that it is actually optimized for better recognition results. With this model, we are able to successfully capture detailed attentional information. Also, the AG-STNs are able to capture attentional information in multiple levels, and different levels of attentional information are complementary to each other in our experiments. A fusion of them brings better results.},
keywords={},
doi={10.1587/transinf.2019EDP7045},
ISSN={1745-1361},
month={December},}
Copy
TY - JOUR
TI - Attention-Guided Spatial Transformer Networks for Fine-Grained Visual Recognition
T2 - IEICE TRANSACTIONS on Information
SP - 2577
EP - 2586
AU - Dichao LIU
AU - Yu WANG
AU - Jien KATO
PY - 2019
DO - 10.1587/transinf.2019EDP7045
JO - IEICE TRANSACTIONS on Information
SN - 1745-1361
VL - E102-D
IS - 12
JA - IEICE TRANSACTIONS on Information
Y1 - December 2019
AB - The aim of this paper is to propose effective attentional regions for fine-grained visual recognition. Based on the Spatial Transformers' capability of spatial manipulation within networks, we propose an extension model, the Attention-Guided Spatial Transformer Networks (AG-STNs). This model can guide the Spatial Transformers with hard-coded attentional regions at first. Then such guidance can be turned off, and the network model will adjust the region learning in terms of the location and scale. Such adjustment is conditioned to the classification loss so that it is actually optimized for better recognition results. With this model, we are able to successfully capture detailed attentional information. Also, the AG-STNs are able to capture attentional information in multiple levels, and different levels of attentional information are complementary to each other in our experiments. A fusion of them brings better results.
ER -