一种图像真伪识别模型训练方法、应用方法和装置
By using a shared feature extraction model to identify the representation information of the image and the image authenticity identification model, and combining loss function optimization, the problem of recognition accuracy under unknown tampering methods and datasets was solved, achieving higher accuracy in image authenticity identification.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- CITY UNIV OF HONG KONG SHENZHEN RES INST
- Filing Date
- 2022-09-21
- Publication Date
- 2026-07-17
AI Technical Summary
Existing authenticity recognition models suffer a significant drop in accuracy when the method of tampering with the image to be identified and the dataset used for tampering are unknown.
The model adopts a preset representation information recognition model and a preset true/false recognition model that share the same feature extraction model. By using sample images, representation information labels and true/false labels during the training process, the loss function value is optimized to improve the recognition accuracy of the model.
In the case of unknown tampering methods and datasets, the accuracy of the authenticity identification model is improved, and it can adaptively extract effective features for image authenticity identification.
Smart Images
Figure CN115496963B_ABST