Biometric Check-in for ATD-ISAP (SmartLINK)
Description
This use case intends to solve the problem of the need for frequent in-person check-ins for participants in the ATD-ISAP program.
Detailed example
There are two outputs related to using ISAP Biometric Monitoring App. Either a participant “passes” (biometric match) or the photo is moved to a “pending review” status. In either scenario, a human can evaluate the response.
AI / analytics pattern
Computer Vision: AI that processes and interprets visual data (e.g., images and videos).
Automation level / stage
c) Deployed – The use case is being actively authorized or utilized to support the functions or mission of an agency.
Expected benefit
ISAP Biometric Monitoring App is a technology option that allows participants to report in using a smartphone. This app verifies a participant’s identity, determine their location, and quickly collect status change information. The app adds functionality not available with telephonic and is less intrusive than a GPS unit. ISAP monitoring app limits in-person interactions of routine check-ins, allowing more time to be allocated to non-compliant participants, complex removal proceedings cases and docket management.
Audit / financial statement impact
Facial verification is only one option for check-in. If the remote check-in fails, either because it was unable to verify the match between the user and their previously taken photo, or because of other potential issues (poor lighting, camera/phone malfunction, etc...), an officer will manually review the check-in photo against the previously taken photos. If that fails, the user can schedule an in-person check-in at their local ERO office. Therefore, the output of AI (facial verification for a remote check-in) is not the primary basis for a decision or action that would affect the individual's rights or safety. It is a convenience to help save time for both the user and officers.
Controls / human review
ATO: Yes; PIA: https://www.dhs.gov/sites/default/files/2023-08/privacy-pia-ice062-atd-august2023.pdf
Data needed
The training process includes datasets from diverse facial images and real-world environments. The images are preprocessed to normalize variables like lighting and facial expressions, making them suitable for facial matching. Data techniques, such as rotation and scaling, are also applied to alleviate the need for additional data collection. The models are trained to extract facial features and to match them accurately.