Professional website of Sébastien Marcel

Talk at ID4Africa 2026

Tue, 12 May 2026 00:00:00 +0000

Benchmarking Multimodal Large Language Models for Face Recognition

Sat, 18 Apr 2026 00:00:00 +0000

Benchmarking Multimodal Large Language Models for Face Recognition

Wed, 01 Apr 2026 00:00:00 +0000

DEMO-AI

Wed, 01 Apr 2026 00:00:00 +0000

Access to factual information is essential for democratic decision-making, public trust, and civic engagement, yet artificial intelligence (AI) enables large-scale creation and dissemination of manipulated content, fabricated narratives, and content amplification that can distort public perception, erode confidence in democratic institutions, and polarize political discourse. These risks threaten to reshape political debates, influence electoral outcomes, and undermine public trust in media sources in Switzerland. Democratic values can be upheld by developing AI tools and governance frameworks to counter disinformation and monitor media framing.

DEMO-AI is an interdisciplinary research project, driving advances in computing to enhance the resilience of democracy, integrating expertise from law, journalism and communication studies, media and information literacy to ensure that AI-supported solutions align with democratic values and regulations. Four project goals include:

AI tools for analyzing news media framing;
AI tools for detecting manipulation of audio-visual media;
legal research on regulatory frameworks for AI and disinformation in Switzerland;
and engaging both the public and professionals in evaluating and testing media tools.

DEMO-AI will produce tools to analyze issue framing and related narratives in Swiss media, facilitate the detection of audio-visual disinformation, and understand legal challenges. These tools will be designed, tested, and refined in collaboration with the general public and professionals, placing their specific needs at the center, thus ensuring real-world applicability. Through societal impact activities, the project extends beyond technology, addressing key challenges across AI, democracy, and policy.

INTERART

Wed, 01 Apr 2026 00:00:00 +0000

The INTERART project brings together the Geneva’s Museum of art and history (MAH), the University of Oxford, the Idiap Research Institute, as well as the School of criminal justice of the University of Lausanne. Together, these institutions are collaborating to uncover the identities of subjects in the MAH’s historical portrait collection, many of whom remain unknown. Notably, the project investigates suspected portraits of Marie-Antoinette, Queen of France, and Marie-Caroline, Queen of Naples, by Jean-Étienne Liotard.

Heterogeneous face recognition is being used to uncover the identities of the sitters. This technology enables a face recognition system to compare faces in diverse media (coloured image, thermal image, drawing, painting). It opens new paths for interpretation and could enable us to reveal the identities of the individuals portrayed.

The project perfectly aligns with Idiap’s vision, demonstrating how artificial intelligence can serve society by unveiling new insights and enriching the disciplines it engages with. It also underscores the wide-ranging applications of AI and the Institute’s cutting-edge expertise.

Supported by the Loterie Romande, the project includes several phases, with an exhibition at the MAH in autumn 2026 and a publication.

Detecting Text Manipulation in Images using Vision Language Models

Sat, 01 Nov 2025 00:00:00 +0000

From Face Recognition to Deception: 20 Years in Biometrics and Synthetic Media (ICCV 2025)

Sun, 19 Oct 2025 00:00:00 +0000

FantasyID: A dataset for detecting digital manipulations of ID-documents

Mon, 01 Sep 2025 00:00:00 +0000

The Invisible Threat: Evaluating the Vulnerability of Cross-Spectral Face Recognition to Presentation Attacks

Mon, 01 Sep 2025 00:00:00 +0000

xEdgeFace: Efficient Cross-Spectral Face Recognition for Edge Devices

Mon, 01 Sep 2025 00:00:00 +0000

ArtFace: Towards Historical Portrait Face Identification via Model Adaptation

Thu, 28 Aug 2025 00:00:00 +0000

DeepID Challenge of Detecting Synthetic Manipulations in ID Documents

Thu, 28 Aug 2025 00:00:00 +0000

EdgeDoc: Hybrid CNN-Transformer Model for Accurate Forgery Detection and Localization in ID Documents

Thu, 28 Aug 2025 00:00:00 +0000

FaceLLM: A Multimodal Large Language Model for Face Understanding

Thu, 28 Aug 2025 00:00:00 +0000

Review of Demographic Fairness in Face Recognition

Thu, 21 Aug 2025 00:00:00 +0000

From Face Recognition to Deception: 20 Years in Biometrics and Synthetic Media (IJCNN 2025)

Thu, 03 Jul 2025 00:00:00 +0000

Synthetic Face Datasets Generation via Latent Space Exploration from Brownian Identity Diffusion

Tue, 01 Jul 2025 00:00:00 +0000

HyperFace: Generating Synthetic Face Recognition Datasets by Exploring Face Embedding Hypersphere

Tue, 01 Apr 2025 00:00:00 +0000

Foundation Models and Biometrics: A Survey and Outlook

Sat, 01 Mar 2025 00:00:00 +0000

Exploring ChatGPT for Face Presentation Attack Detection in Zero and Few-Shot In-Context Learning

Sat, 01 Feb 2025 00:00:00 +0000

AugGen: Synthetic Augmentation using Diffusion Models Can Improve Recognition

Wed, 01 Jan 2025 00:00:00 +0000

CERTAIN

Wed, 01 Jan 2025 00:00:00 +0000

CERTAIN focuses on the ethical and regulatory transparency of AI systems, with the goal of helping organizations assess and improve compliance in a practical and technically grounded way.

The project delivers guidelines and tools to support regulatory compliance, assess data quality, measure bias in datasets, and protect privacy. It aligns closely with broader work on trustworthy biometrics and responsible AI by combining technical evaluation with legal, ethical, and operational considerations.

Digi2Real: Bridging the Realism Gap in Synthetic Data Face Recognition via Foundation Models

Sun, 01 Dec 2024 00:00:00 +0000

An overview of 20 years of research in biometrics and recent work (EPFL)

Wed, 22 May 2024 00:00:00 +0000

ROSALIND

Thu, 01 Feb 2024 00:00:00 +0000

ROSALIND focuses on defending digital identity systems against malicious AI-generated face images and manipulated identity documents.

The project combines two complementary goals: developing robust anti-fraud defenses for document images and selfie-videos, and using generative AI to improve the robustness and balance of authentication algorithms. It sits at the intersection of biometric security, digital identity, and applied trustworthy AI.

ROSALIND is a strong example of translational biometrics research with direct relevance to real-world identity verification.

CARMEN

Mon, 01 Jan 2024 00:00:00 +0000

CARMEN develops biometric solutions for non-stop border control for both pedestrians and vehicles in uncontrolled environments.

The project addresses the practical difficulties of “on-the-move” biometrics, including lower-quality live data, lack of time to read ePassports, and real operational constraints outside controlled indoor checkpoints. It aims to make biometric border technologies more accurate, reliable, and deployable in realistic large-scale scenarios.

Example Talk: Recent Work

Mon, 01 Jan 2024 00:00:00 +0000

Example Talk

Dr. Alex Johnson · Meta AI

Research Overview

Multimodal LLMs
Efficient training
Responsible AI

Code & Math

def score(x: int) -> int:
    return x * x

$$ E = mc^2 $$

Dual Column Layout

Left Column

Point A
Point B
Point C

Right Column

Detail 1
Detail 2
Detail 3

Main Content (2/3 width)

This column takes up twice the space of the right column.

def example():
    return "code works too"

Note
Key points in smaller column

Image + Text Layout

Results

95% accuracy
10x faster inference
Lower memory usage

Breakthrough!

Speaker Notes

Press S to open presenter view with notes!

This slide has hidden speaker notes below.

Note:

This is a speaker note (only visible in presenter view)
Press S key to open presenter console
Perfect for remembering key talking points
Can include reminders, timing, references
Supports Markdown formatting too!

Progressive Reveals

Content appears step-by-step:

First point appears Then the second point Finally the conclusion This one can be highlighted!

Note: Use fragments to control pacing and maintain audience attention. Each fragment appears on click.

Diagrams with Mermaid

graph LR A[Research Question] --> B{Hypothesis} B -->|Valid| C[Experiment] B -->|Invalid| D[Revise] C --> E[Analyze Data] E --> F{Significant?} F -->|Yes| G[Publish] F -->|No| D

Perfect for: Workflows, architectures, processes

Note: Mermaid diagrams are created from simple text. They’re version-controllable and edit anywhere!

Research Results

Model	Accuracy	Speed	Memory
Baseline	87.3%	1.0x	2GB
Ours (v1)	92.1%	1.5x	1.8GB
Ours (v2)	95.8%	2.3x	1.2GB

Key Finding: 8.5% improvement over baseline with 40% memory reduction

Note: Tables are perfect for comparative results. Markdown tables are simple and version-control friendly.

Custom Backgrounds

This slide has a blue background!

You can customize:

Background colors
Background images
Gradients
Videos (yes, really!)

Use {{< slide background-color="#hex" >}}

There’s more content below! ⬇️

Press the Down Arrow to see substeps.

Note: This demonstrates Reveal.js’s vertical slide feature. Great for optional details or deep dives.

Substep 1: Details

This is additional content in a vertical stack.

Navigate down for more, or right to skip to next topic →

Substep 2: More Details

Even more detailed information.

Press Up Arrow to go back, or Right Arrow to continue.

Citations & Quotes

“The best way to predict the future is to invent it.”

— Alan Kay

Or reference research:

Recent work by Smith et al. (2024) demonstrates that Markdown-based slides improve reproducibility by 78% compared to proprietary formats¹.

Media: YouTube Videos

Note: Embed YouTube videos with just the video ID. Perfect for demos, tutorials, or interviews.

Media: All Options

Embed various media types with simple shortcodes:

YouTube: {{< youtube VIDEO_ID >}}
Bilibili: {{< bilibili id="BV1..." >}}
Local videos: {{< video src="file.mp4" controls="yes" >}}
Audio: {{< audio src="file.mp3" >}}

Perfect for demos, interviews, tutorials, or podcasts!

Note: All media types work seamlessly in slides. Just use the appropriate shortcode.

Interactive Elements

Try these keyboard shortcuts:

→ ← : Navigate slides
↓ ↑ : Vertical navigation
S : Speaker notes
F : Fullscreen
O : Overview mode
/ : Search
ESC : Exit modes

Hidden Slide Demo (Inline Comment)

This slide is hidden using the  comment method.

Perfect for:

Speaker-only content
Backup slides
Work-in-progress content

Note: This slide won’t appear in the presentation but remains in source for reference.

Thanks

Questions?

🌐 Website:
🐦 X/Twitter:
💬 Discord:
⭐ GitHub:

All slides created with Markdown • No vendor lock-in • Edit anywhere

Note: Thank you for your attention! Feel free to reach out with questions or contributions.

🎨 Branding Your Slides

Add your identity to every slide with simple configuration!

What you can add:

Element	Position Options
Logo	top-left, top-right, bottom-left, bottom-right
Title	Same as above
Author	Same as above
Footer Text	Same + bottom-center

Edit the branding: section in your slide’s front matter (top of file).

📁 Adding Your Logo

Place your logo in assets/media/ folder
Use SVG format for best results (auto-adapts to any theme!)
Add to front matter:

branding:
  logo:
    filename: "your-logo.svg"  # Must be in assets/media/
    position: "top-right"
    width: "60px"

Tip: SVGs with fill="currentColor" automatically match theme colors!

📝 Title & Author Overlays

Show presentation title and/or author on every slide:

branding:
  title:
    show: true
    position: "bottom-left"
    text: "Short Title"  # Optional: override long page title
  
  author:
    show: true
    position: "bottom-right"

Author is auto-detected from page front matter (author: or authors:).

📄 Footer Text

Add copyright, conference name, or any persistent text:

branding:
  footer:
    text: "© 2024 Your Name · ICML 2024"
    position: "bottom-center"

Tip: Supports Markdown! Use [Link](url) for clickable links.

🔇 Hiding Branding Per-Slide

Sometimes you want a clean slide (title slides, full-screen images).

Add this comment at the start of your slide content:

<!-- no-branding -->
## My Clean Slide

Content here...

☝️ This slide uses  — notice no logo or overlays!

🔇 Selective Hiding

Hide just the header (logo + title):

<!-- no-header -->

Or just the footer (author + footer text):

<!-- no-footer -->

☝️ This slide uses  — footer still visible below!

✅ Quick Reference

Comment	Hides
`<!-- no-branding -->`	Everything (logo, title, author, footer)
`<!-- no-header -->`	Logo + Title overlay
`<!-- no-footer -->`	Author + Footer text

☝️ This slide uses  — logo still visible above!

🚀 Get Started

Copy this example’s front matter as a starting point
Replace logo with yours in assets/media/
Customize positions and text
Use  for special slides

Pro tip: Set site-wide defaults in config/_default/params.yaml under slides.branding!

Smith, J. et al. (2024). Open Science Presentations. Nature Methods. ↩︎

PopEye

Mon, 01 Jan 2024 00:00:00 +0000

PopEye develops robust privacy-preserving biometric technologies for passenger identification and verification at EU external borders.

The project addresses operational constraints such as open-air conditions, night-time acquisition, time pressure, and large-scale throughput, while emphasizing privacy-preserving design. Its goal is to improve the accuracy, reliability, and usability of biometric recognition in demanding border-management scenarios.

PopEye represents a current strand of work where biometric performance, privacy, and deployment realism must all be addressed together.

Biometrics Security

Fri, 01 Dec 2023 00:00:00 +0000

Experience

Tue, 24 Oct 2023 00:00:00 +0000

Face Presentation Attack Detection (EPFL)

Mon, 15 May 2023 00:00:00 +0000

Spoofing and Anti-Spoofing in Biometrics

Wed, 07 Dec 2022 00:00:00 +0000

Spoofing and Anti-Spoofing in Biometrics

Sat, 01 Oct 2022 00:00:00 +0000

SAFER

Tue, 01 Mar 2022 00:00:00 +0000

SAFER addresses fairness and ethics in face recognition.

The project investigates how to assess and reduce unfair performance differences across demographic groups, with work spanning both training-time and scoring-time strategies. It also explores the role of synthetic and diverse datasets in improving the responsible development of face recognition systems.

SAFER reflects a broader commitment to trustworthy biometrics by combining technical performance with fairness, transparency, and responsible deployment.

BATL

Sun, 01 Jan 2017 00:00:00 +0000

BATL contributed to biometric anti-spoofing research in the context of the IARPA ODIN program, which aimed to strengthen biometric systems against known and unknown presentation attacks.

Within this line of work, research focused on robust face presentation-attack detection, anomaly detection, multi-channel sensing, and the creation of challenging datasets and protocols for evaluating spoof resilience. At Idiap, this effort is closely associated with work on multi-channel face anti-spoofing and datasets such as HQ-WMCA, supporting reproducible research on secure biometric acquisition.

BEAT

Thu, 01 Mar 2012 00:00:00 +0000

BEAT (Biometrics Evaluation and Testing) was a European FP7 project focused on creating an open and reproducible framework for the evaluation of biometric technologies.

The project addressed three complementary goals: transparent benchmarking of biometric systems, vulnerability analysis, and support for standardized evaluation procedures. It contributed to reproducible research practices in biometrics and helped establish evaluation workflows that were both rigorous and operationally relevant.

This project also supported the broader vision of open platforms and tools for the community, connecting research, benchmarking, and technology assessment.

TABULA RASA

Fri, 01 Jan 2010 00:00:00 +0000

TABULA RASA was a major European research project dedicated to understanding and mitigating spoofing attacks against biometric systems.

The project investigated the vulnerability of biometric modalities such as face and fingerprint to direct attacks and developed methods to detect and counter such threats. It played an important role in shaping the modern field of presentation-attack detection and helped establish anti-spoofing as a core topic in trustworthy biometrics.

This project remains a key milestone in the evolution of biometric security research and in the translation of anti-spoofing knowledge to practical evaluation settings.

MOBIO

Tue, 01 Jan 2008 00:00:00 +0000

MOBIO focused on mobile biometrics under realistic usage conditions, combining face and voice for authentication in noisy and unconstrained environments.

The project investigated robust face localisation, speech segmentation, video-based face authentication, speaker authentication, multimodal fusion, and unsupervised model adaptation over time. It also contributed a widely used multimodal database collected across multiple countries and sites, helping establish strong evaluation benchmarks for mobile biometric research.

Professional website of Sébastien Marcel

Talk at ID4Africa 2026

Benchmarking Multimodal Large Language Models for Face Recognition

Benchmarking Multimodal Large Language Models for Face Recognition

DEMO-AI

INTERART

Detecting Text Manipulation in Images using Vision Language Models

From Face Recognition to Deception: 20 Years in Biometrics and Synthetic Media (ICCV 2025)

FantasyID: A dataset for detecting digital manipulations of ID-documents

The Invisible Threat: Evaluating the Vulnerability of Cross-Spectral Face Recognition to Presentation Attacks

xEdgeFace: Efficient Cross-Spectral Face Recognition for Edge Devices

ArtFace: Towards Historical Portrait Face Identification via Model Adaptation

DeepID Challenge of Detecting Synthetic Manipulations in ID Documents

EdgeDoc: Hybrid CNN-Transformer Model for Accurate Forgery Detection and Localization in ID Documents

FaceLLM: A Multimodal Large Language Model for Face Understanding

Review of Demographic Fairness in Face Recognition

From Face Recognition to Deception: 20 Years in Biometrics and Synthetic Media (IJCNN 2025)

Synthetic Face Datasets Generation via Latent Space Exploration from Brownian Identity Diffusion

HyperFace: Generating Synthetic Face Recognition Datasets by Exploring Face Embedding Hypersphere

Foundation Models and Biometrics: A Survey and Outlook

Exploring ChatGPT for Face Presentation Attack Detection in Zero and Few-Shot In-Context Learning

AugGen: Synthetic Augmentation using Diffusion Models Can Improve Recognition

CERTAIN

Digi2Real: Bridging the Realism Gap in Synthetic Data Face Recognition via Foundation Models

An overview of 20 years of research in biometrics and recent work (EPFL)

ROSALIND

CARMEN

Example Talk: Recent Work

Example Talk

Dr. Alex Johnson · Meta AI

Research Overview

Code & Math

Dual Column Layout

Left Column

Right Column

Main Content (2/3 width)

Sidebar (1/3 width)

Image + Text Layout

Results

Speaker Notes

Progressive Reveals

Diagrams with Mermaid

Research Results

Custom Backgrounds

Vertical Navigation

Substep 1: Details

Substep 2: More Details

Citations & Quotes

Media: YouTube Videos

Media: All Options

Interactive Elements

Hidden Slide Demo (Inline Comment)

Thanks

Questions?

🎨 Branding Your Slides

📁 Adding Your Logo

📝 Title & Author Overlays

📄 Footer Text

🔇 Hiding Branding Per-Slide

🔇 Selective Hiding

✅ Quick Reference

🚀 Get Started

PopEye

Biometrics Security

Experience

Face Presentation Attack Detection (EPFL)

Spoofing and Anti-Spoofing in Biometrics

Spoofing and Anti-Spoofing in Biometrics

SAFER

BATL

BEAT

TABULA RASA

MOBIO