- Cross-medium Watermarking
Cultural and creative digital media works are susceptible to piracy infringements during storage, transmission, and interaction, such as print scanning, print shooting, and screen shooting. This project studies the key technologies and application of image/video-oriented cross-medium media bridge based on deep learning frameworks focusing on the secure interaction between physical substance and Internet digital media. This work is supported by National Natural Science Foundation of China, Key Project U22A2030.
- Image Forensics of Operator Chains
Actual image forgeries may contain multiple operations, which jointly constitute an image operator chain in a specific order. Up to now, current passive image forensics remains at the stage of detecting some specific image manipulation or some type of image operations, which cannot meet the deep requirements for practical forensics scenarios. This project researches on the passive forensics of image operator chains for processing history estimation.
- Deepfake Detection
Deepfake detection is the defense demand of information and public opinion war in the era of AIGC. This project aims at the challenges of continuous update of Deepfake videos and weakening of its forensic clues, developing cognitive countermeasure technology in the video field to proactively cope with the offensive and defensive confrontation in the cognitive domain of cyberspace.