iTC Meeting Minutes 2022 05 17 - biometricITC/cPP-biometrics GitHub Wiki

Agenda

Attendees

  • Brian Wood

  • Clare Parran

  • Greg Fiumara

  • Naruki Kai

  • Greg Ott

  • Ahmad Dahari Jarno

  • Jim Arnold

  • Douglas Kalmus

Record of Decisions

  • None

Action Items

  • Naruki will provide feedback on Issue #371

  • Brian will create pull requests for some issues

Minutes

The call started with a review of the task list. Brian will add a new item for the MDF issues.

Brian then noted that the MDF comment period closed last week, so that should be published in the not too distant future. Given the issues that were reported, the CMFA work will pause to focus on the biometrics requirements as they will be used in evaluations soon.

The call then moved on to the Supporting Doc Issues that were raised by evaluators from a lab review of the requirements.

The first issue discussed was Issue #367 about making sure that tests are practical and cost effective. The PR #379 adds a statement in the BIOSD for this, but it isn’t clear if this is needed here (it could also go in the README). This will be reviewed and discussed on the next call.

The next issue discussed was Issue #368 (and Issue #369) about how the evaluator should understand the quality requirements given the labs are unlikely to be experts in biometric systems. The intent for the requirement is not that the evaluator necessarily evaluates the details of the quality metrics (to prove they are good or bad), but that there are quality metrics and they are being used. Based on this, it is felt that some tweaks should be made to the requirements, and Brian will propose some initial changes for review on the next call for further discussion.

The next topic was Issue #370 about how to do the testing on the template data. The proposed PR #380 for this was discussed and felt it was not exactly the right approach. After discussion it was felt that a "live" test during the biometric transaction probably wasn’t useful, and was definitely going to cause problems for producing results. Brian will propose an update to reflect this difference. Additionally a note about not specifically needing to look for the entire template as a block should be provided so the search is more flexible. It was noted that the template header information would probably be a good thing to look for. It was also pointed out that it may be beneficial to have some sort of test about the collected sample as opposed to just the template, but it wasn’t clear how this would work (given the nature of different sensors).

The last topic of review was Issue #371 about the TSFI testing. Naruki said he would provide some information about what the intent of this test is supposed to be before the next call.

The call ended at 10:59am EDT.

⚠️ **GitHub.com Fallback** ⚠️