LivDet – Face 2024
Algorithms and Systems
Liveness Detection Competition
2nd Edition
Competition Description
LivDet-Face 2024 is the second competition in the LivDet-Face series offering (a) an independent assessment of the current state of the art in Face Presentation Attack Detection algorithms and (b) an evaluation protocol, including datasets of spoof and live face images and videos, that can be followed by researchers after the competition is closed and compare their solutions with LivDet-Face winners and baselines.
LivDet-Face 2024 has been included in the official IJCB 2024 competition list.
This competition will have two parts:
Part 1: “Algorithms” will involve the evaluation of the PAD solutions submitted by the competitors and performed by the organizers on a sequestered dataset unknown to the competitors.
Part 2: “Systems” will involve the systematic testing of submitted face recognition systems based on physical artifacts presented to the sensors.
Competitors can participate in one, two, or both parts. For each part, a separate winner demonstrating the best performance will be announced. Metrics recommended by ISO/IEC 30107-1:2016 will be used to assess the submissions.
The process to enter the LivDet-Face 2024 competition will be as follows:
Step 1. Register: Select an anonymous participant for the publication.
Step 2. Submit algorithms for evaluation(one for image and one for video).
Step 3. Processing: Execute testing protocol.
Step 4. Performance will be assessed.
Step 5. Results will be announced and one winner will be selected.
Step 6: Publication-Teams who submit 3 best-performing methods will be invited to submit a joint paper summarizing the competition and describing briefly the PAD methods.
Important Dates
Mar 15th, 2024
Registration and PAD submission are now open.
Sample data is available upon registration.
May 15th, 2024
Submission deadline
Participants’ deadline for submission of PAD solutions.
Optional: short descriptions of the methods of non-anonymous participants who want to co-author the IJCB paper.
June 26th, 2024
Result Completed
The best teams will be invited to co-author the IJCB paper submission.
How to Participate
To participate in Part 1 — “Algorithms”:
Follow these instructions to submit your executables on or before the due date of May 15th, 2024 (End of Day, Anywhere on Earth). Provide the following data (in the email) with your submission:
Submitter name:
Affiliation:
Email address:
Phone number:
Mailing address:
Acronym or short name of the team/solution: (if you want to stay anonymous in the competition and all resulting publications, say “Anonymous”).
Submission Format Requirements: Algorithm
Two PAD software must be provided as a console application executable for image and video attack testing.
The application should be in the format of a .exe file that can be directly opened and executed in a Windows/Linux terminal or command prompt.
The software must accept input files via the command line to perform processing. A readme file on
Upon processing, the software should generate the test score and output the results to a specified .txt file.
By adhering to the specified format, you will enable efficient testing and streamlined scoring procedures for your PAD software submissions.
To participate in Part 2 — “Systems”:
A data-sharing agreement forms shared and has to be signed by the participants/organizers to exchange sample data.
The LivDet-Face team will process the executable application file on the sequestered datasets (however with possibly different attack types) and CANNOT be used in any way for training your algorithms.
You will need to have your Face recognition system submitted to Clarkson University, NY, USA. Please contact us at your earliest convenience at livdetface24@gmail.com to arrange the logistics of this testing.
Submission request Part 2: Systems.
Submission can be one of the following.
a) Full hardware capture system + PAD software
b) Software/app which can run on a mobile device or computer.
Discuss compatibility with available devices with organizers
c) Submission should capture a face image and provide a log file with PAD scores.
The README file should cover the following:
Detailed instructions on running the algorithm or system.
Clear guidelines on specifying input and the corresponding output.
Any additional instructions or parameters required to initiate the testing process and obtain the test score results.
Your attention to these submission guidelines and testing instructions is greatly appreciated, as they will ensure the efficient and accurate evaluation of your PAD algorithm/system.
If you have any inquiries or require further clarification, please feel free to contact us via email.
FAQ
- Is a Win32 application required? “Other formats are possible with approval from the LivDet team”.
- Should it be a single executable? “it can be delivered as a zip file containing additional files or folders (such as models, configurations, etc.)”.
- What is the format of the software to be submitted? Is there a standard API? “The application should be in the format of a .exe file that can be directly opened and executed in a Windows terminal or command prompt.”
- Is there a description of the types of presentation attacks we would expect to encounter? “Images and video attack forms”.
- what should the score look like? “PAD score should range from 0 to 100 threshold set a 50 threshold”. 50-100 as live, and 0-49 as spoof.
- If we submit, will the results be public like NIST FRTE or will there be a codename approach similar to the DHS Rally? “Participants can decide to use a code name or submit anonymously” Choice anonymous or not – Participants will receive evaluated results and they can desire. Non-anonymous – Coauthor the publication if the author or system performs amongst the best.
- Metrics evaluated? “BPCER, APCER, ACER, and IAPAR with FRR, if matching and PAD analysis is carried out.”
Results and IJCB Paper Submission
Winners:
Each part (1, and 2) will have its separate winner. Metrics recommended by ISO/IEC 30107-1:2016 will be used to assess the submissions. For Parts 1 and 2 the 50.0 threshold will be used to calculate APCER and BPCER error rates.
Competition Results – Part 1
The full breakdown of the competition results will be relayed to competitors and made available in the IJCB 2024 paper.
IJCB Paper Submission:
At least all winners (from all categories) will be invited to co-author the IJCB paper submission. Depending on the number and quality of submissions (e.g., when the performance between the winners and runner-ups is small), we will extend this invitation to the authors of other well-performing algorithms and systems, to make the resulting paper useful for the community and the participants.
Have additional questions?
We’re here to help.