The original paper is in English. Non-English content has been machine-translated and may contain typographical errors or mistranslations. ex. Some numerals are expressed as "XNUMX".
Copyrights notice
The original paper is in English. Non-English content has been machine-translated and may contain typographical errors or mistranslations. Copyrights notice
En tirant parti de l’apprentissage profond et de l’apprentissage par renforcement, ADNet (Action Decision Network) surpasse les autres approches. Cependant, sa vitesse et ses performances sont encore limitées par des facteurs tels qu'une estimation peu fiable du score de confiance et des actions historiques redondantes. Pour remédier aux limitations ci-dessus, une approche plus rapide et plus précise nommée Faster-ADNet est proposée dans cet article. En optimisant le processus de suivi via un réseau de réidentification de statut, l'approche proposée est plus efficace et 6 fois plus rapide qu'ADNet. Dans le même temps, la précision et la stabilité sont améliorées par la suppression des actions historiques. Les expériences démontrent les avantages de Faster-ADNet.
Tiansa ZHANG
Beijing Institute of Technology,Chinese Academy of Sciences
Chunlei HUO
Chinese Academy of Sciences
Zhiqiang ZHOU
Beijing Institute of Technology
Bo WANG
Beijing Institute 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.
Copier
Tiansa ZHANG, Chunlei HUO, Zhiqiang ZHOU, Bo WANG, "Faster-ADNet for Visual Tracking" in IEICE TRANSACTIONS on Information,
vol. E102-D, no. 3, pp. 684-687, March 2019, doi: 10.1587/transinf.2018EDL8214.
Abstract: By taking advantages of deep learning and reinforcement learning, ADNet (Action Decision Network) outperforms other approaches. However, its speed and performance are still limited by factors such as unreliable confidence score estimation and redundant historical actions. To address the above limitations, a faster and more accurate approach named Faster-ADNet is proposed in this paper. By optimizing the tracking process via a status re-identification network, the proposed approach is more efficient and 6 times faster than ADNet. At the same time, the accuracy and stability are enhanced by historical actions removal. Experiments demonstrate the advantages of Faster-ADNet.
URL: https://global.ieice.org/en_transactions/information/10.1587/transinf.2018EDL8214/_p
Copier
@ARTICLE{e102-d_3_684,
author={Tiansa ZHANG, Chunlei HUO, Zhiqiang ZHOU, Bo WANG, },
journal={IEICE TRANSACTIONS on Information},
title={Faster-ADNet for Visual Tracking},
year={2019},
volume={E102-D},
number={3},
pages={684-687},
abstract={By taking advantages of deep learning and reinforcement learning, ADNet (Action Decision Network) outperforms other approaches. However, its speed and performance are still limited by factors such as unreliable confidence score estimation and redundant historical actions. To address the above limitations, a faster and more accurate approach named Faster-ADNet is proposed in this paper. By optimizing the tracking process via a status re-identification network, the proposed approach is more efficient and 6 times faster than ADNet. At the same time, the accuracy and stability are enhanced by historical actions removal. Experiments demonstrate the advantages of Faster-ADNet.},
keywords={},
doi={10.1587/transinf.2018EDL8214},
ISSN={1745-1361},
month={March},}
Copier
TY - JOUR
TI - Faster-ADNet for Visual Tracking
T2 - IEICE TRANSACTIONS on Information
SP - 684
EP - 687
AU - Tiansa ZHANG
AU - Chunlei HUO
AU - Zhiqiang ZHOU
AU - Bo WANG
PY - 2019
DO - 10.1587/transinf.2018EDL8214
JO - IEICE TRANSACTIONS on Information
SN - 1745-1361
VL - E102-D
IS - 3
JA - IEICE TRANSACTIONS on Information
Y1 - March 2019
AB - By taking advantages of deep learning and reinforcement learning, ADNet (Action Decision Network) outperforms other approaches. However, its speed and performance are still limited by factors such as unreliable confidence score estimation and redundant historical actions. To address the above limitations, a faster and more accurate approach named Faster-ADNet is proposed in this paper. By optimizing the tracking process via a status re-identification network, the proposed approach is more efficient and 6 times faster than ADNet. At the same time, the accuracy and stability are enhanced by historical actions removal. Experiments demonstrate the advantages of Faster-ADNet.
ER -