Title:
METHOD AND APPARATUS FOR GENERATING FEATURE EXTRACTION NETWORK, AND DEVICE AND COMPUTER-READABLE MEDIUM
Document Type and Number:
WIPO Patent Application WO/2022/012179
Kind Code:
A1
Abstract:
A method and apparatus for generating a feature extraction network, and an electronic device and a computer-readable medium. The method comprises: respectively inputting a first sample image and a second sample image into a feature extraction network to obtain a first sample feature map and a second sample feature map, wherein the second sample image is obtained by performing an affine transformation on the first sample image (201); performing an affine transformation on the first sample feature map to obtain a first sample affine transformation feature map (202); with regard to a first vector and a second vector at the same position in the first sample affine transformation feature map and the second sample feature map, determining loss values of the first vector and the second vector on the basis of a pre-set loss function (203); and training the feature extraction network on the basis of the loss values (204). By means of the method, optimization of the training of a feature extraction network is realized, and the features extracted from an image subjected to an affine transformation are similar to the features extracted from an original image.
More Like This:
Inventors:
HE YI (CN)
WANG CHANGHU (CN)
WANG CHANGHU (CN)
Application Number:
PCT/CN2021/096145
Publication Date:
January 20, 2022
Filing Date:
May 26, 2021
Export Citation:
Assignee:
BEIJING BYTEDANCE NETWORK TECH CO LTD (CN)
International Classes:
G06T3/00
Foreign References:
| CN111915480A | 2020-11-10 | |||
| CN111340013A | 2020-06-26 | |||
| CN111382793A | 2020-07-07 | |||
| CN111382727A | 2020-07-07 | |||
| CN102231191A | 2011-11-02 | |||
| JP2007133840A | 2007-05-31 |
Attorney, Agent or Firm:
TAHOTA LAW FIRM (CN)
Download PDF: