MODEL TRAINING METHOD, WATERMARK RESTORATION METHOD, AND RELATED DEVICE

WIPO Patent Application WO/2023/246403

A1

The present application provides a model training method, a watermark restoration method, and a related device. The training method comprises: obtaining watermark pattern information and background pattern information; performing combination according to the watermark pattern information and the background pattern information to generate a watermark image set; compressing a watermark image in the watermark image set to obtain a corresponding watermark compressed image, and respectively cropping the watermark image and the corresponding watermark compressed image to obtain a plurality of watermark image blocks corresponding to the watermark images and a plurality of watermark compressed image blocks corresponding to the positions of the watermark image blocks; using the watermark compressed image blocks as training samples, using the watermark image blocks corresponding to the positions of the watermark compressed image blocks as sample tags, and combining the training samples and the sample tags corresponding to training samples to generate a training data set; and constructing a neural network model, and calling the training data set to train the neural network model by means a convergence acceleration algorithm to obtain a neural network model meeting a training end condition as a watermark restoration model.

SUN WEIDI (CN)
GUO FENG (CN)
SU XIAODONG (CN)

PCT/CN2023/095675

December 28, 2023

May 23, 2023

Click for automatic bibliography generation  

BEIJING VOLCANO ENGINE TECH CO LTD (CN)

G06T1/00

CN111953993A	2020-11-17
CN105992072A	2016-10-05
CN111223034A	2020-06-02
KR20120055070A	2012-05-31
US20190213705A1	2019-07-11

SINOTALEN ATTORNEYS AT LAW (CN)

View/Download PDF      PDF Help

Previous Patent: MODEL TRAINING METHOD, WATERMARK TEXT RECOGNITION METHOD, AND RELATED DEVICE

Next Patent: COMMUNICATION METHOD AND APPARATUS