Domain incremental learning in forgery detection

Digital image forgery has become a worldwide pandemic, with many forms of forgery (e.g., insurance fraud, fake news, identity theft) negatively affecting our life. This effect could be attributed to the accessible costs of mobile phones and digital cameras, which has led to an exponential proliferation of digital images, and the availability of many image editing tools that allow easy manipulation of images, resulting in high-quality forgeries. In this project, we intend to leverage the powerful representation capabilities of deep models to address the problem of forgery detection in realistic and challenging scenarios, such as incrementally learning across continuously evolving domains.

Faculty Supervisor:

Jose Dolz

Student:

Partner:

Thales Recherche et Technologie

Discipline:

Computer science

Sector:

Manufacturing; Professional, scientific and technical services

University:

École de technologie supérieure

Program: