TY - GEN
T1 - DEEPMIX
T2 - 2021 IEEE International Conference on Multimedia and Expo, ICME 2021
AU - Cheng, Ziyi
AU - Ren, Xuhong
AU - Juefei-Xu, Felix
AU - Xue, Wanli
AU - Guo, Qing
AU - Ma, Lei
AU - Zhao, Jianjun
N1 - Funding Information:
Acknowledgement. This work was supported in part by the National Natural Science Foundation of China under Grant 61906135, 62020106004, and 92048301, Tianjin Science and Technology Plan Project under Grant 20JCQNJC01350, JSPS KAKENHI Grant No.20H04168, 19K24348, 19H04086, and JST-Mirai Program Grant No.JPMJMI18BB and JPMJMI20B8, Japan, and Natural Science Foundation of Tianjin under Grant KJZ40420200017. This work was also supported by the Canada CIFAR AI program.
Publisher Copyright:
© 2021 IEEE
PY - 2021
Y1 - 2021
N2 - Online updating of the object model via samples from historical frames is of great importance for accurate visual object tracking. Recent works mainly focus on constructing effective and efficient updating methods while neglecting the training samples for learning discriminative object models, which is also a key part of a learning problem. In this paper, we propose the DeepMix that takes historical samples' embeddings as input and generates augmented embeddings online, enhancing the state-of-the-art online learning methods for visual object tracking. More specifically, we first propose the online data augmentation for tracking that online augments the historical samples through object-aware filtering. Then, we propose MixNet which is an offline trained network for performing online data augmentation within one-step, enhancing the tracking accuracy while preserving high speeds of the state-of-the-art online learning methods. The extensive experiments on three different tracking frameworks, i.e., DiMP, DSiam, and SiamRPN++, and three large-scale and challenging datasets, i.e., OTB-2015, LaSOT, and VOT, demonstrate the effectiveness and advantages of the proposed method.
AB - Online updating of the object model via samples from historical frames is of great importance for accurate visual object tracking. Recent works mainly focus on constructing effective and efficient updating methods while neglecting the training samples for learning discriminative object models, which is also a key part of a learning problem. In this paper, we propose the DeepMix that takes historical samples' embeddings as input and generates augmented embeddings online, enhancing the state-of-the-art online learning methods for visual object tracking. More specifically, we first propose the online data augmentation for tracking that online augments the historical samples through object-aware filtering. Then, we propose MixNet which is an offline trained network for performing online data augmentation within one-step, enhancing the tracking accuracy while preserving high speeds of the state-of-the-art online learning methods. The extensive experiments on three different tracking frameworks, i.e., DiMP, DSiam, and SiamRPN++, and three large-scale and challenging datasets, i.e., OTB-2015, LaSOT, and VOT, demonstrate the effectiveness and advantages of the proposed method.
UR - http://www.scopus.com/inward/record.url?scp=85125339031&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85125339031&partnerID=8YFLogxK
U2 - 10.1109/ICME51207.2021.9428185
DO - 10.1109/ICME51207.2021.9428185
M3 - Conference contribution
AN - SCOPUS:85125339031
T3 - Proceedings - IEEE International Conference on Multimedia and Expo
BT - 2021 IEEE International Conference on Multimedia and Expo, ICME 2021
PB - IEEE Computer Society
Y2 - 5 July 2021 through 9 July 2021
ER -