TY - GEN
T1 - AFSN
T2 - 8th International Conference on Big Data Computing and Communications, BigCom 2022
AU - Feng, Xuyang
AU - Guo, Bin
AU - Qiu, Chen
AU - Ren, Haoyang
AU - Yu, Zhiwen
N1 - Publisher Copyright:
© 2022 IEEE.
PY - 2022
Y1 - 2022
N2 - Viewing real-time human synthesis results on smart-phones provides users a convenient interactive experience. How-ever, most human synthesis solutions relay on high-performance servers and lack interaction with users. In order to reduce traffic between smartphones and cloud, this paper introduces a novel adaptive frame selection network (AFSN) on mobile devices. Besides, we propose a reinforcement-learning based agent that selects the image frames according to the pose change, that reduces communications and the amount of calculations. Based upon AFSN, we build a multi-task human synthesis system, that can deal with background changing, pose transfer, virtual try-on, and other generate tasks in a unified framework. Experimental results on one real-world dataset demonstrate that our method can reduce the number of frames to be processed by more than 50%.
AB - Viewing real-time human synthesis results on smart-phones provides users a convenient interactive experience. How-ever, most human synthesis solutions relay on high-performance servers and lack interaction with users. In order to reduce traffic between smartphones and cloud, this paper introduces a novel adaptive frame selection network (AFSN) on mobile devices. Besides, we propose a reinforcement-learning based agent that selects the image frames according to the pose change, that reduces communications and the amount of calculations. Based upon AFSN, we build a multi-task human synthesis system, that can deal with background changing, pose transfer, virtual try-on, and other generate tasks in a unified framework. Experimental results on one real-world dataset demonstrate that our method can reduce the number of frames to be processed by more than 50%.
KW - cloud-device collaborative
KW - human image synthesis
KW - reinforcement learning
UR - https://www.scopus.com/pages/publications/85151496794
U2 - 10.1109/BigCom57025.2022.00036
DO - 10.1109/BigCom57025.2022.00036
M3 - 会议稿件
AN - SCOPUS:85151496794
T3 - Proceedings - 2022 8th International Conference on Big Data Computing and Communications, BigCom 2022
SP - 224
EP - 232
BT - Proceedings - 2022 8th International Conference on Big Data Computing and Communications, BigCom 2022
PB - Institute of Electrical and Electronics Engineers Inc.
Y2 - 6 August 2022 through 7 August 2022
ER -