Skip to main content

image Manipulation Attack Resolving Solutions

Periodic Reporting for period 1 - iMARS (image Manipulation Attack Resolving Solutions)

Reporting period: 2020-09-01 to 2021-12-31

iMARS is a European H2020 project, funded under Grant Agreement ID 883356. It started in September 2020 and will continue until August 2024.
iMARS purpose is to develop solutions for the fight against identity document fraud.
Fraudulent identity documents are used for almost all organised crime activities (drugs, people trafficking, terrorist activities, etc.). They are therefore a major problem for the society.
Some fraudulent identity documents are regularly issued by government entities: they carry false information because the breeder documents or enrolment data on which they rely are false. This is the case with identity documents showing a fake face image, made from two different people, and looking equally like both of them. The process used to obtain such fake images is called morphing.
iMARS considers three use-cases in the life cycle of identity documents:
• Document issuing
• Document control at the border
• Forensic investigation
For all three use-cases, iMARS focuses on two sets of technical tools useful in the fight against fraudulent documents.
• MAD (Manipulation Attack Detection) tools to detect image manipulation and especially morphing.
• DVFD (Document Validation and Fraud Detection) tools to detect if identity documents are fraudulent or not.
iMARS also develops a number of critical enablers for the uptake of its technological tools and further research in the same field:
• Legal studies to define the legal context of the usage of iMARS tools,
• Social acceptance studies,
• Ethical studies related to the iMARS project itself and the operational usage of its technologies,
• Study and proposed improvements of the processes corresponding to the use-cases, and especially the enrolment phase in identity document issuing,
• Skill assessment and training of the key personnel involved in the use-cases of iMARS,
• User experience study of the technical tools involving the end-users,
• Development of a testing database for MAD tools that will be maintained 20 years after iMARS,
• Standards in the field of face images, related to quality and manipulation and presentation attack detection, including metrics.
The list below presents the work performed and the results obtained in iMARS in period 1 (Sept 2020 to Dec 2021).
• Management rules definition (complete), and actual management (on going).
• Use-case analysis (complete)
• Requirement definition for the tools of iMARS (MAD and DVFD) (complete)
• iMARS testing methodology and test plans definition (complete)
• iMARS test database gathering (on going): the high quality image database is now available and can be used for MAD tools benchmarking
• Analysis of legal requirements related to iMARS, and presentation of them to technical people (complete).
• Legal framework for the collection of biometric data for long term research (complete)
• First report by an independent ethical advisor, also public and re-usable (complete)
• Analysis of societal acceptability of iMARS technology using a survey (on going)
• Risk analysis of the current enrolment processes and design of secure enrolment processes (complete). From this study, a policy brief will be extracted and forwarded to relevant European authorities
• Study of how automatic face recognition systems handle manipulated images in the document control use case (on going)
• Study of the ability of human experts to detect morphs (on going). A large-scale study was completed (490 people). Its results will soon be published, showing the impact of training, the impact of the initial working position, and how humans compare with automatic morphing attack detection tools.
• Communication infrastructure of iMARS (complete) and actual communication (on going)
• Publication (on going): 14 publications completed or prepared
• Presentation, workshop organisation and interviews (on going; 33 completed in Period 1)
• Standardization (on going)
o on face quality: the iMARS team input has been presented to SC37 Working Group 3
o on presentation attack detection
• In DVFD, design of three innovative ways of detecting forged Identity documents. Design of a security framework to protect personal information in DVFD.
• Study of the evolution of morphing techniques in the coming years: this is a critical input in iMARS, because the Morphing Attack Detection techniques developed in iMARS will not be already obsolete when the project stops.
• Proof that morphing techniques have a visible impact on images that can be highlighted by different image processing techniques.
• In the field of Manipulation Attack Detection, which is the main purpose of iMARS, design and test of new MAD algorithms, both for Single MAD (S-MAD, where only one suspicious image is available, e.g. from a passport), and Differential MAD (D-MAD, when the passport image and a live image of the person presenting it are available). Results of international evaluations were published just after the end of period 1.
During period 1, the work performed in iMARS has already led to above state of the art results in the field of
• Morphing techniques knowledge
• DVFD solutions
• MAD algorithms, for both Single Mad and Differential MAD.
• Availability in Europe of an international MAD algorithm evaluation platform
• Knowledge of human ability to detect morphed face images
• Standards
• Knowledge of the legal framework applicable to the development of solutions to fight against document fraud

Because of these early encouraging results, it is expected that in the coming periods iMARS will be able to reach its objectives, namely
• Deliver image manipulation and morphing detection software, validated in relevant environment, that will improve the practitioners’ fraud detection capacity
• Deliver efficient and mobile solutions for ID document verification, checking document validity, detecting and counteracting document fraud, validated in relevant environment
• Deliver face image quality assessment software that will predict recognition performance, validated in relevant environment
• Deliver iMARS technologies fully compliant with EU Legal-Ethical-Privacy rules
• Ensure the uptake of the iMARS technologies beyond the end of the project
We will have at mid-project, as initially planned, clear, realistic and reproducible benchmarks against which to assess progress, to possibly stop the project if progress is not deemed sufficient. However, we do not anticipate at this stage that results will be disappointing and that it will be required to stop the project.
It is also expected that iMARS will be able to reach its expected impact, as defined before project starts:
• Plan to provide confidence in the take up of project results after the completion of the project.
• Evidence based knowledge, and developments performing beyond the current state of the art and leading quickly to innovation.
• Technical and operational guidelines, recommendations and best practices set in the EUROSUR handbook
• Implementation of solutions resulting from the legislative initiative in the "Smart Borders" package
• Implementation of the actions identified by the EU Strategy and Action Plan for customs risk management