For this second version, with the support of Facephi company, HDA, and Fraunhopher, we will use a custom platform based on or similar to EvalAI to handle submissions and evaluations online directly for each participant team. 

• All the teams must submit a docker image with an API that receives an image file as input in JPEG, PNG, or TIFF format and returns a continuous float value in the range [0.0, 1.0], where 0.0 represents total confidence that the image is a Presentation Attack and 1.0 represents total confidence that the image is Bona fide. 

• Each team's docker image implementation must include all specific processes, such as segmentation and alignment.

• We will provide a GitHub repository with an example and instructions on building the image and preparing it for submission. [Link Github]. 

• Each submission will be evaluated in an offline environment, so the docker image should include all necessary packages and data (weights, configuration files, etc.) to work without external connectivity.

• All participants need to submit the registration form. Once accepted, we will provide a user for the evaluation platform. [Link Platform].

For the performance evaluation, we will follow the metrics recommended by ISO/IEC 30107-3, which will be used to assess all submissions: the Equal Error Rate, Bonafide Presentation Classification Error Rate (BPCER), and Attack Presentation Classification Error Rate (APCER). 

We will report the three working points recommended by ISO/IEC 30107-3, which are the BPCER value when the APCER score is fixed at 10% (BPCER10), 5% (BPCER20), and 1% (BPCER100).