Visualization of Facial Trajectories for Fast Adjudication in Watchlist Screening

Public security organizations have deployed video surveillance at many security checkpoints, at e.g., airport inspection lanes and border crossings. Face4 Systems provides solutions for watchlist screening based on state-of-the-art commercial technologies. Although automated still-to-video FR is a valuable function in video surveillance, screening for threatening individuals is largely limited to the FR abilities of a human operator. FR systems suffer from the effects of pose, illumination, resolution, blur, etc. With current technologies, watchlist screening may yield several false matches and nonmatches errors. Fast adjudication is therefore needed to limit the number of costly false matches. The objective of this project is to develop and evaluate algorithms that are suitable to support fast adjudication of faces match reports (produced by an automated FR system) in real-world watchlist screening applications. Given a set of facial captures (trajectory) linked face match, two specific types of visualisation functions are considered to facilitate adjudication – (1) quality assessment for analysis, ranking and selection of probe ROIs; and (2) super-resolution for enhancement appearance of probe face captures. A specialized target-based evaluation methodology will be developed to assess the benefit of proposed visualisation functions for fast adjudication using publicly-available and operational videos collected with Face4 Systems.

Faculty Supervisor:

Éric Granger

Student:

Ibtihel Amara, Farshad Nourbakhsh, Roghaiyeh Soleimani

Partner:

Face4 Systems

Discipline:

Engineering - computer / electrical

Sector:

Information and communications technologies

University:

École de technologie supérieure