CVPR 2024 Biometrics Workshop
In conjunction with the IEEE/CVF Conference on Computer Vision and Pattern Recognition · June 17, 2024
Call for Papers
In conjunction with CVPR 2024, this workshop focuses on advanced biometrics research and the fusion of biometric samples, sensing modes, and modalities.
With a security-conscious society, biometrics-based authentication and identification have become central to many important applications because biometrics can provide accurate and reliable identification. Biometrics research and technology continue to mature rapidly, driven by industrial and government needs and supported by industrial and government funding. As the number and types of biometric architectures and sensors increase, the need to disseminate research results increases as well. This workshop is positioned at the frontier of biometrics research and showcases excellent and advanced work from academic and private research organizations as well as government labs.
Many applications require a level of accuracy that is not feasible with a single biometric. Fusing multiple biometrics can improve population coverage for people who cannot provide a single biometric and can improve system security against spoof attacks. The workshop addresses research issues in different modes and levels of fusion, with the goal of improving biometric performance. Theoretical studies on sensor fusion techniques for biometric authentication, recognition, and performance are encouraged.
Topics of interest:
- Sensing: intensity, depth, thermal, pressure, time-series, and exotic sensing.
- Face, finger, ear, eye, iris, retina, vein pattern, palm, gait, foot, and exotic biometrics.
- Biometric template computation, feature extraction, and matching.
- Data and performance baselines.
- Evolution of standards, competitions, and organized challenge problems.
- Score-level, decision-level, and feature-level integration.
- Architectures for integration and evidence integration.
- Fusion-based identification techniques.
- Normalization techniques involved in fusion techniques.
- Machine learning techniques in biometric fusion.
- Public databases and score files in multi-biometrics.
- Application-dependent personalization of multi-biometric systems.
- Theoretical models for integration.
- Performance modeling, prediction, and evaluation of multi-biometric systems.
- Security improvement assessment for multi-biometric systems.
Overall Meeting Sponsors
Organizing Committee
General Chair
Bir Bhanu
University of California, Riverside
bbhanu@ucr.edu
Program Chair
Ajay Kumar
The Hong Kong Polytechnic University
Ajay.Kumar@polyu.edu.hk
Program Committee
- Naser Damer, Fraunhofer IGD, Germany
- Hazim Ekenel, Istanbul Technical University, Turkey
- Hugo J. Escalante, INAOE, Puebla, Mexico
- Patrick Flynn, Notre Dame University, USA
- Ruber Hernández García, Universidad Católica del Maule, Chile
- Manuel Günther, University of Zurich, Switzerland
- Jiankun Hu, University of New South Wales, Australia
- Koichi Ito, Tohoku University, Japan
- Zhen Lei, NLPR, CASIA, China
- Manhua Liu, Shanghai Jiao Tong University, China
- Domingo Mery, Universidad Católica de Chile, Chile
- Aythami Morales, Universidad Autónoma de Madrid, Spain
- Tetsushi Ohki, Shizuoka University, Japan
- Hugo Proença, University of Beira Interior, Portugal
- Raghavendra Ramachandra, NTNU, Norway
- William Schwartz, Federal University of Minas Gerais, Brazil
- Kar-Ann Toh, Yonsei University, South Korea
- Jun Wan, NLPR, CASIA, China
- Shiqi Yu, Southern University of Science and Technology, China
Important Dates
Paper submission
March 4, 2024 · 11:59 PM PST
Decisions
April 7, 2024
Camera-ready papers due
April 13, 2024
Workshop date
June 17, 2024
Technical Program
CVPR 2024 Biometrics Workshop · June 17, 2024
| Code | Session, paper, or event |
|---|---|
| S1 | Face Recognition 9:00–10:15 AM · Session Chair: Raghavendra Ramachandra (Norwegian University of Science and Technology, Norway) |
| FIQA-FAS: Face Image Quality Assessment Based Face Anti-Spoofing 9:00–9:25 · Ya Chi Liang, Min-Xuan Qiu, Shang-Hong Lai (National Tsing Hua University, Taiwan) |
|
| GraFIQs: Face Image Quality Assessment Using Gradient Magnitudes 9:25–9:50 · Jan Niklas Kolf (Fraunhofer IGD, Germany), Naser Damer (Fraunhofer IGD and TU Darmstadt), Fadi Boutros (Fraunhofer IGD, Germany) |
|
| Can the accuracy bias by facial hairstyle be reduced through balancing the training data? 9:50–10:15 · Kagan Ozturk, Haiyu Wu, Kevin W. Bowyer (University of Notre Dame, USA) |
|
| Coffee Break 10:15–10:30 AM |
|
| T1 | Biometric Sample Quality: the Recognition Model Perspective 10:30–11:30 AM · Naser Damer, Fraunhofer IGD, Germany |
| S2 | Biometric Synthesis and Recognition 11:30 AM–12:30 PM · Session Chair: Rui Zhao (Amazon, USA) |
| One Embedding to Predict Them All: Visible and Thermal Universal Face Representations for Soft Biometric Estimation via Vision Transformers 11:30–11:45 · Nélida Mirabet-Herranz, Chiara Galdi, Jean-Luc Dugelay (EURECOM, France) |
|
| Confidence-Aware RGB-D Face Recognition via Virtual Depth Synthesis 11:45 AM–12:00 PM · Zijian Chen, Mei Wang, Weihong Deng, Hongzhi Shi, Dongchao Wen, Yingjie Zhang, Xingchen Cui, Jian Zhao (Inspur and collaborators, China) |
|
| TattTRN: Template Reconstruction Network for Tattoo Retrieval 12:00–12:15 · Lazaro Janier Gonzalez-Soler (Hochschule Darmstadt, Germany), Maciej Salwowski (Technical University of Denmark), Christian Rathgeb, Daniel Fischer (Hochschule Darmstadt, Germany) |
|
| 3D Kinematics Estimation from Video with a Biomechanical Model and Synthetic Training Data 12:15–12:30 · Zhi-Yi Lin, Bofan Lyu, Judith Cueto Fernandez, Eline van der Kruk, Ajay Seth, Xucong Zhang (Delft University of Technology, The Netherlands) |
|
| Lunch Break 12:30–2:00 PM |
|
| T2 | Human Recognition at a Distance in Video 2:00–3:00 PM · Bir Bhanu, University of California, Riverside |
| Coffee Break 3:00–3:15 PM |
|
| S3 | Adversarial Attacks and Detection 3:15–4:30 PM · Session Chair: Ruben Vera-Rodriguez (Universidad Autónoma de Madrid, Spain) |
| Outsmarting Biometric Imposters: Enhancing Iris-Recognition System Security through Physical Adversarial Example Generation and PAD Fine-Tuning 3:15–3:40 · Yuka Ogino, Kazuya Kakizaki, Takahiro Toizumi, Atsushi Ito (NEC and University of Tsukuba, Japan) |
|
| Adversarial Identity Injection for Semantic Face Image Synthesis 3:40–4:05 · Zijian Chen, Mei Wang, Weihong Deng, Hongzhi Shi, Dongchao Wen, Yingjie Zhang, Xingchen Cui, Jian Zhao (Inspur and collaborators, China) |
|
| Generalized Single-Image-Based Morphing Attack Detection Using Deep Representations from Vision Transformer 4:05–4:30 · Haoyu Zhang, Raghavendra Ramachandra, Kiran Raja, Christoph Busch (NTNU, Norway) |
|
| T3 | Recent Advances in Privacy-Preserving Biometrics Authentication 4:30–5:30 PM · Jiankun Hu, University of New South Wales, Australia |
| Awards, Valedictory and Closing Remarks 5:30–5:45 PM · Bir Bhanu, University of California, Riverside, USA |
S1 and S3 papers have 25-minute oral presentation slots: 18 minutes for presentation, 5 minutes for questions and answers, and 2 minutes for setup. S2 papers have 15-minute slots: 8 minutes for presentation, 5 minutes for questions and answers, and 2 minutes for setup.
Invited Talk and Panel Session
Biometric Sample Quality: the Recognition Model Perspective
Speaker: Naser Damer, Fraunhofer Institute for Computer Graphics Research IGD, Germany
Abstract: Biometric sample quality measures the utility of a sample to the recognition algorithm. Its relevance to different biometric-system processes has attracted the attention of multiple stakeholders, as shown by the NIST Face Analysis Technology Evaluation (FATE) Quality evaluation and the evolving ISO standardization effort, including ISO/IEC 29794-1. Using face recognition as an example, this talk presents modern face-recognition models, their training, and template inference, then discusses recognition-model behavior that can indicate sample quality. Observations from inference and training can be translated into quality scores that indicate biometric-sample utility.
Human Recognition at a Distance in Video
Speaker: Bir Bhanu, University of California, Riverside
Abstract: Most biometric systems for human recognition require physical contact with, or close proximity to, a cooperative subject. More challenging is reliably recognizing non-cooperative individuals at a distance from arbitrary angles under changing real-world conditions. Gait, face, and body biometrics can be captured from a distance using video cameras mounted on land or aerial platforms, but imaging distortions, range, limited frames, arbitrary pose, occlusions, air turbulence, and changing clothes create technical challenges. This talk presents 2D and 3D representations and methods for robust video-based human recognition at a distance in the wild.
Recent Advances in Privacy-Preserving Biometrics Authentication
Speaker: Jiankun Hu, University of New South Wales, Australia
Abstract: Biometrics can authenticate genuine users and is used in applications such as border control and digital access control, but biometric data is privacy-sensitive and regulated in many countries. This talk introduces major research in privacy-preserving biometric authentication, including infinite-to-one mapping-based cancelable biometric template design, Attack via Record Multiplicity (ARM), ARM-attack-resilient cancelable biometric designs, hill-climbing attacks on biometric templates, resilient cancelable deep-learning models, and Biometrics-Based Authenticated Key Exchange Protocol with Multi-Factor Fuzzy Extractor.
Paper Submission
All submissions are subject to a double-blind review process. Author names, affiliations, email addresses, personal acknowledgements, and similar identifying information must be removed from the paper. Papers may be up to 8 pages, excluding references, during the review process.
- All papers must use the IEEE format and the provided LaTeX or MS-Word templates.
- Submitted papers must not have been published, accepted, or be under review elsewhere.
- Accepted papers must be presented by one of the authors, and at least one author must register at the standard registration rate before the paper-registration deadline. An accepted paper not presented by a registered co-author will be recommended for withdrawal from IEEE Xplore.
- Authors must transfer copyright to IEEE for papers published in the conference proceedings when submitting the camera-ready version. If a paper is withdrawn, copyright reverts to the authors.
Submission system: CMT for CVPRW Biometrics 2024
Accepted Papers
- 3D Kinematics Estimation from Video with a Biomechanical Model and Synthetic Training Data
- Outsmarting Biometric Imposters: Enhancing Iris-Recognition System Security through Physical Adversarial Example Generation and PAD Fine-Tuning
- FIQA-FAS: Face Image Quality Assessment Based Face Anti-Spoofing
- Adversarial Identity Injection for Semantic Face Image Synthesis
- Confidence-Aware RGB-D Face Recognition via Virtual Depth Synthesis
- Face Image Quality Assessment Using Gradient Magnitudes
- One Embedding to Predict Them All: Visible and Thermal Universal Face Representations for Soft Biometric Estimation via Vision Transformers
- Generalized Single-Image-Based Morphing Attack Detection Using Deep Representations from Vision Transformer
- Can the accuracy bias by facial hairstyle be reduced through balancing the training data?
- TattTRN: Template Reconstruction Network for Tattoo Retrieval
Reviews were available through CMT after April 7, 2024. Camera-ready submission details and instructions were sent by email after that date.
Challenges
Awards
Workshop Venue
Seattle, USA
The CVPR 2024 Biometrics Workshop was held in Seattle, Washington, USA.