Face Presentation Attack Database for Smart Glass Capture (SmaGla-FPAD)

Most existing face PAD benchmarks are acquired with fixed cameras or handheld devices and therefore do not fully capture the variability introduced by first person wearable acquisition. Smart glass capture is more challenging because it combines ego motion, viewpoint variation, background clutter, changing subject to sensor distance, and less controlled illumination. To study PAD under these conditions, we introduce SmaGla-FPAD, a smart glass based face presentation attack database collected using Ray-Ban Meta Wayfarer Smart Glasses with a front facing RGB camera supporting 12 MP image capture and video recording up to 3K resolution. Figure 1 illustrates the acquisition process.

The database is collected in three settings, namely Indoor, Outdoor, and in-car, chosen to reflect capture scenarios relevant to border control and mobile identity verification. The Indoor setting is acquired in a corridor under controlled lighting, whereas the Outdoor and in-car settings are recorded under natural daylight and therefore exhibit stronger variation in illumination, shadow, and background content. In Indoor and Outdoor settings, subjects walk naturally towards the smart glass wearer from an initial distance of approximately 6 m, which introduces progressive changes in scale and pose. In the in-car setting, the subject remains seated while the smart glass device records from a distance of 1 to 2 m. To increase scene diversity, the in-car subset is collected across 10 vehicle types of different sizes.

SmaGla-FPAD contains two classes, namely Bona fide and Presentation attack. In total, the database comprises 297,000 sample images, including 225,000 bona fide samples and 72,000 presentation attack samples. The bona fide subset contains recordings from 50 subjects aged between 18 and 40 years, including 18 male and 32 female participants. For the Indoor and Outdoor settings, each subject is recorded over multiple sessions spanning four months, with 5 to 8 videos per subject in each setting, and each sequence lasting between 30 and 50 seconds. In the in-car setting, recordings are acquired from both left and right sides of the vehicle.

The presentation attack subset is generated using four custom built silicone masks, comprising two male and two female masks, worn by 15 subjects. Silicone masks are used because they represent a realistic three dimensional presentation attack instrument with high visual fidelity and substantial surface structure. To increase acquisition realism, accessories such as wigs, hoodies, and eyeglasses are incorporated during recording. Two attack conditions are considered: in the first, the attacker wears a silicone mask together with eyeglasses, a wig, and a hoodie; in the second, the attacker wears the same accessories except for eyeglasses. For each condition, 5 to 12 videos of 40 to 60 seconds are recorded over multiple sessions within 30 days. Table 1 summarises the dataset composition, and Figure 2 shows representative examples from the three settings.

Copyright of SmaGla-FPAD :

The researcher can access the  Face Presentation Attack Database for Smart Glass Capture (SmaGla-FPAD) by following the procedure mentioned: 

• Researchers are required to send the request to use MMPA by sending a copy of the license agreement, completely filled, to vetrekarnarayan@unigoa.ac.in with the subject line "License agreement for Face Presentation Attack Database for Smart Glass Capture (SmaGla-FPAD)". 

Note: The license agreement has to be signed by the researcher or supervisor and the signature of a legal authority on behalf of the institution, such as the Head of the institution or Registrar, along with the institutional seal (Rubber Stamp).  The license agreement should be on the researchers institutional letterhead. 

The request will not be considered if due procedures are not followed.  

• Every request to avail SmaGla-FPAD will be placed before the Ethical Committee of the Institute for approval, and the confirmation of the request will be sent via email. Further, all the instructions related to access to the database will be detailed in the email.

This database is available only for research and educational purposes and not for any commercial use. Further, the use of the database is only for one year from the date of signing the agreement by the researcher. Beyond the allocated one year for the usage of data, the applicant has to apply again following the due procedure.  

The database is available only for research and educational purposes. All the rights of the Face Presentation Attack Database for Smart Glass Capture (SmaGla-FPAD) are reserved, and commercial use/distribution of this database is strictly prohibited. All the technical reports and papers that report experimental results from this database should provide due acknowledgement and reference. If you use the database in any publications or reports, you must refer to the following paper:  

• Raghavendra Ramachandra, Narayan Vetrekar, Neha Naik, and Rajendra Gad, Decoupled Representation Learning for Face Presentation Attack Detection under Smart Glass Capture, IJCB 2026, Rome, Italy.(Accepted). (code)