Title:
WEAKLY PAIRED IMAGE STYLE TRANSFER METHOD BASED ON POSE SELF-SUPERVISED GENERATIVE ADVERSARIAL NETWORK
Document Type and Number:
WIPO Patent Application WO/2023/284070
Kind Code:
A1
Abstract:
A weakly paired image style transfer method based on a pose self-supervised generative adversarial network, relating to the field of image processing. The method is suitable for style transfer of weakly paired images, different styles of pictures having certain overlap are used to perform model training of an adversarial neural network, so that the model is insensitive to poses and focuses on style learning, and therefore, in an actual application process, a source style can be converted into a target style, but a pose is kept unchanged. In addition, in the model training process of the adversarial neural network, a differentiable pose solver capable of estimating a relative pose of any two images is introduced, a phase correlation algorithm is optimized to be differentiable, and the phase correlation algorithm is embedded into an end-to-end learning network framework to achieve pose estimation. Style transfer of weakly paired data can be achieved, and thus support is provided for robot self-positioning technology.
More Like This:
| JP2006129401 | IMAGE PROCESSING APPARATUS |
| WO/2018/113339 | PROJECTION IMAGE CONSTRUCTION METHOD AND DEVICE |
| JP2004248290 | VIGNETTING COMPENSATION |
Inventors:
WANG YUE (CN)
CHEN ZEXI (CN)
GUO JIAXIN (CN)
XU XUECHENG (CN)
WANG YUNKAI (CN)
XIONG RONG (CN)
CHEN ZEXI (CN)
GUO JIAXIN (CN)
XU XUECHENG (CN)
WANG YUNKAI (CN)
XIONG RONG (CN)
Application Number:
PCT/CN2021/113982
Publication Date:
January 19, 2023
Filing Date:
August 23, 2021
Export Citation:
Assignee:
UNIV ZHEJIANG (CN)
International Classes:
G06T3/00; G06N3/08
Foreign References:
| CN111932438A | 2020-11-13 | |||
| CN111783525A | 2020-10-16 | |||
| US20200151559A1 | 2020-05-14 |
Other References:
ZEXI CHEN; JIAXIN GUO; XUECHENG XU; YUNKAI WANG; YUE WANG; RONG XIONG: "PREGAN: Pose Randomization and Estimation for Weakly Paired Image Style Translation", ARXIV.ORG, CORNELL UNIVERSITY LIBRARY, 201 OLIN LIBRARY CORNELL UNIVERSITY ITHACA, NY 14853, 17 January 2021 (2021-01-17), 201 Olin Library Cornell University Ithaca, NY 14853 , XP081882365, DOI: 10.1109/LRA.2021.3061359
Attorney, Agent or Firm:
HANGZHOU QIUSHI PATENT OFFICE CO., LTD. (CN)
Download PDF: