Iris Liveness Database
- Datasets from previous LivDet-Iris competitions will be shared for training.
- A small subset of samples from the test dataset will be shared to provide an overview of the spoof types
- Spoofs: Five categories (Paper prints, Patterned Contacts, Cadaver Eyes, Electronic Display, Fake/Prosthetic Eyes and Printed with Add ons), some unknown attacks which are not present in training set
- Live data: Combination of same sensor and cross-sensor data
- For more detailed information on database check Dataset.
Iris Baseline PAD Algorithm
An open source algorithm is shared with the participants as an additional resource. Participants may choose to use this if they want. The baseline results from this algorithm will be published.
Step 1. Register: Select an anonymous participant for the publication.
Step 2. Sign and submit training database license agreement – optional; relevant documents will be shared via email upon registration.
Step 3. Access Databases: On submitting licence agreement, information will be shared for downloading training datasets.
Step 4. Submit algorithms for evaluation. For detailed instruction click here.
Step 5. Processing: Execute the testing protocol on the BEAT platform if submission is via BEAT platform.
Step 6. Performance will be assessed.
Step 7. Results will be announced and one winner will be selected.
Step 8: 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 .